Not in the top 1,000 by GitHub stars
SkillMedev/skills
The open Skill Me catalog — every hosted skill as a portable, MIT-licensed SKILL.md
History
Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.
View weekly changes
skills/1on1-agenda/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/ab-test-analyzer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/abandoned-cart-sequence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/academic-essay/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/accessibility-audit/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/ad-creative-testing/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/aeo-answer-blockifier/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/agency-monthly-report/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/agency-positioning/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/agency-qbr-upsell/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/agent-orchestration/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/alert-tuning/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/amazon-listing-optimizer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/animation-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/annual-plan/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/api-client-generator/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/api-design/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/api-versioning-strategist/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/apollo-prospecting/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/app-store-copy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/app-store-release-prep/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/async-communication/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/audience-targeting/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/big-purchase-decision/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/bloodwork-explainer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/board-management/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/book-proposal/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/brand-guidelines/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/brand-naming/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/budget-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/budget-pacing/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/budget-vs-actual/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/build-on-agent-native/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/business-rule-extractor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/buyer-nurture-sequence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/buying-signal-tracker/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/caching-strategy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/candidate-outreach-personalizer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cap-table-manager/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/captions-from-transcript/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/carousel-scripter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/case-study-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cash-flow-forecast/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/category-page-copywriter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/causal-inference/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/changelog-generator/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/changelog-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/channel-strategy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/characterization-test-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/churn-reduction/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/circuit-breaker-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/citation-tracker/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/claims-verifier/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/clickhouse-analytics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/client-churn-save/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/client-comms-cadence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/client-onboarding-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/client-status-reporting/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/clinical-summary/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/closer-sales-script/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cma-narrative-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/code-review-checklist/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cold-email-craft/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cold-email-deliverability/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/color-accessibility/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/color-palette-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/community-engagement-playbook/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/comparison-page-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/competitive-intelligence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/competitive-moat/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/component-api-design/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/conflict-resolution/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/connection-pool-tuner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/content-brief/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/content-refresh-auditor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/content-repurposing/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/contract-test-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/core-four-lead-engine/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/course-outline-architect/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/course-sales-page/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cover-letter-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/coverage-gap-finder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/creator-content-calendar/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/crisis-comms-external/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/cross-platform-reformatter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/csat-root-cause/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/curriculum-mapper/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/customer-analytics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/customer-success-qbr/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/dark-mode-design/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/data-drift-monitor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/data-quality/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/data-room-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/data-story/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/data-table-design/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/database-schema/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/dead-code-eliminator/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/debt-payoff-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/decision-making/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/deep-research/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/deep-work-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/demo-script/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/dependency-risk-audit/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/design-critique/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/design-handoff-doc/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/design-qa-checklist/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/design-token-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/differentiated-instruction/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/discovery-call-prep/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/docker-compose-wizard/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/document-template-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/donor-communications/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/due-diligence-checklist/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/e2e-scenario-author/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/economic-analysis/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/eda-playbook/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/email-deliverability/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/email-drip-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/email-newsletter-pro/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/email-triage/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/emergency-fund-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/employment-contract/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/eng-status-rollup/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/engagement-reply-drafter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/env-doctor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/error-budget-policy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/error-handling/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/error-message-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/escalation-summary/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/executive-summary/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/expansion-revenue/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/expense-policy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/experiment-tracking/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/expert-interview/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/explain-plan-reader/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fact-checker/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/feature-flags/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/feature-store-design/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/feedback-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fiction-scene-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/finance-fx/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/financial-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/financial-statement-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fitness-program/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/flaky-test-detangler/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/flutter-widget-architect/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/formatted-report-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fpa-model/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/framework-upgrader/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/freelance-positioning/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fun-content/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fundraise-closing-mechanics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fundraise-readiness-audit/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fundraise-team-hiring/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fundraising-narrative/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/fundraising-stage-selector/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/funnel-analysis/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/geo-places/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/ghostwriter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/git-commit-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/github-actions/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/go-to-market-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/goals-accountability/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/google-business-profile-optimizer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/government-open-data/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/grand-slam-offer-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/grant-loi-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/grant-proposal-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/grant-prospect-researcher/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/graphql-schema/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/growth-accounting/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/growth-model/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/gtd-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/gym-meta-ads-funnel/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/gym-money-model/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/gym-pricing-and-guarantees/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/gym-transformation-challenge/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/habit-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/hashtag-strategist/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/help-documentation/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/hiring-a-players-and-sops/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/hiring-pipeline/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/hiring-scorecard/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/icon-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/icp-persona-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/idempotency-enforcer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/image-license-rights/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/impact-report-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/inbox-zero/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/index-advisor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/injury-prehab/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/instagram-post-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/instagram-trend-scout/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/internal-linking-mapper/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/interview-debrief-synthesizer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/interview-guide-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/interview-question-kit/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/interview-synthesis/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/investment-basics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/investor-pipeline-crm/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/investor-targeting/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/investor-update-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/jd-bias-scrubber/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/jetpack-compose-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/jira-ticket-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/job-application/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/job-description-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/job-interview-prep/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/journal-framework/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/jtbd-extractor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/kafka-pipelines/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/kb-article-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/keyword-cluster-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/kinetic-typography/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/kpi-scoreboard-and-cadence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/kubernetes-basics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/landing-page-copy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/landing-page-cro/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/language-learning/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/language-reference/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/language-version-migrator/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/launch-day-runbook/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/launch-email-sequence/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/launch-plan-sequencer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/lead-enrichment/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/lesson-plan-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/life-coach/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/lifecycle-journey-map/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/linear-workflow/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/linkedin-post-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/listing-description-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/literature-review/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/llm-evaluation/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/load-testing/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/local-referral-engine/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/local-seo-playbook/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/local-unit-economics/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/localization-guide/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/logo-brief-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/longevity-protocol/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/m-and-a-analysis/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/manuscript-reviser/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/market-sizing/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/marketing-attribution/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mcp-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/meal-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/media-monitor/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/meeting-agenda/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/meeting-notes-to-actions/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/messaging-hierarchy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/meta-title-optimizer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/microservices/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/migration-safety-checker/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/ml-feature-engineering/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mobile-offline-sync/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mobile-perf-profiler/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mobility-routine/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mock-stub-designer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/model-card-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/model-evaluation-report/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mongodb-expert/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/monolith-decomposer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/monorepo-expert/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/month-end-close/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/moodboard-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/motion-color-and-light/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/motion-design-principles/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/motion-spec/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mutation-test-runner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/mutual-action-plan/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/n-plus-one-hunter/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/narration-script/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/networking-system/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/next-on-vercel-perf/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/nextjs-app-router/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/no-show-reduction/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/nonprofit-board-pack/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/notion-database/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/nutrition-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/oauth-flow/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Network or broad shell → After Network or broad shell
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Network or broad shell
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/objection-handler/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/objection-handling-and-speed-to-lead/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/observability-stack/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/offer-rejection-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/okr-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/onboarding-copy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/onboarding-plan-builder/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/oncall-handoff/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/one-pager-designer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/op-ed-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/open-house-follow-up/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/outreach-sequence-designer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/pagination-and-sync-engineer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/paid-acquisition-audit/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/pandas-expert/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/partition-planner/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/partnership-strategy/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/patent-prior-art/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/performance-review-writer/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
skills/pii-scrubber/SKILL.md
Week of Nov 2, 2026 · +0 / −0 lines
Before Instructions only → After Instructions only
97a13607d669 → 97a13607d669
Week of Oct 26, 2026 · Line counts unavailable
Before Not checked yet → After Instructions only
First observation → 97a13607d669
No earlier version was observed for comparison.
24GitHub stars
4stars in 7 days
482skills in this repository
3commits in 30 days
Heat
4.1- Star growth, last 7 daysNot available
- Star growth, last 24 hoursNot available
- Rank movement, last 7 daysNot available
- Commits, last 30 daysNot available
- Accounts asking about it through MCP, last 7 daysNot available
- Hot votes this weekNot available
Heat never changes the order by stars, and sponsorship never touches it. How heat works
Stars over time
30-day change not yet available: not enough history.
Individual skills in this repository
- 1on1-agenda
skills/1on1-agenda - ab-test-analyzer
skills/ab-test-analyzer - abandoned-cart-sequence
skills/abandoned-cart-sequence - academic-essay
skills/academic-essay - accessibility-audit
skills/accessibility-audit - ad-creative-testing
skills/ad-creative-testing - aeo-answer-blockifier
skills/aeo-answer-blockifier - agency-monthly-report
skills/agency-monthly-report - agency-positioning
skills/agency-positioning - agency-qbr-upsell
skills/agency-qbr-upsell - agent-orchestration
skills/agent-orchestration - alert-tuning
skills/alert-tuning - amazon-listing-optimizer
skills/amazon-listing-optimizer - animation-system
skills/animation-system - annual-plan
skills/annual-plan - api-client-generator
skills/api-client-generator - api-design
skills/api-design - api-versioning-strategist
skills/api-versioning-strategist - apollo-prospecting
skills/apollo-prospecting - app-store-copy
skills/app-store-copy - app-store-release-prep
skills/app-store-release-prep - async-communication
skills/async-communication - audience-targeting
skills/audience-targeting - big-purchase-decision
skills/big-purchase-decision - bloodwork-explainer
skills/bloodwork-explainer - board-management
skills/board-management - book-proposal
skills/book-proposal - brand-guidelines
skills/brand-guidelines - brand-naming
skills/brand-naming - budget-builder
skills/budget-builder - budget-pacing
skills/budget-pacing - budget-vs-actual
skills/budget-vs-actual - build-on-agent-native
skills/build-on-agent-native - business-rule-extractor
skills/business-rule-extractor - buyer-nurture-sequence
skills/buyer-nurture-sequence - buying-signal-tracker
skills/buying-signal-tracker - caching-strategy
skills/caching-strategy - candidate-outreach-personalizer
skills/candidate-outreach-personalizer - cap-table-manager
skills/cap-table-manager - captions-from-transcript
skills/captions-from-transcript - carousel-scripter
skills/carousel-scripter - case-study-builder
skills/case-study-builder - cash-flow-forecast
skills/cash-flow-forecast - category-page-copywriter
skills/category-page-copywriter - causal-inference
skills/causal-inference - changelog-generator
skills/changelog-generator - changelog-writer
skills/changelog-writer - channel-strategy
skills/channel-strategy - characterization-test-writer
skills/characterization-test-writer - churn-reduction
skills/churn-reduction - circuit-breaker-builder
skills/circuit-breaker-builder - citation-tracker
skills/citation-tracker - claims-verifier
skills/claims-verifier - clickhouse-analytics
skills/clickhouse-analytics - client-churn-save
skills/client-churn-save - client-comms-cadence
skills/client-comms-cadence - client-onboarding-system
skills/client-onboarding-system - client-status-reporting
skills/client-status-reporting - clinical-summary
skills/clinical-summary - closer-sales-script
skills/closer-sales-script - cma-narrative-builder
skills/cma-narrative-builder - code-review-checklist
skills/code-review-checklist - cold-email-craft
skills/cold-email-craft - cold-email-deliverability
skills/cold-email-deliverability - color-accessibility
skills/color-accessibility - color-palette-builder
skills/color-palette-builder - community-engagement-playbook
skills/community-engagement-playbook - comparison-page-builder
skills/comparison-page-builder - competitive-intelligence
skills/competitive-intelligence - competitive-moat
skills/competitive-moat - component-api-design
skills/component-api-design - conflict-resolution
skills/conflict-resolution - connection-pool-tuner
skills/connection-pool-tuner - content-brief
skills/content-brief - content-refresh-auditor
skills/content-refresh-auditor - content-repurposing
skills/content-repurposing - contract-test-writer
skills/contract-test-writer - core-four-lead-engine
skills/core-four-lead-engine - course-outline-architect
skills/course-outline-architect - course-sales-page
skills/course-sales-page - cover-letter-writer
skills/cover-letter-writer - coverage-gap-finder
skills/coverage-gap-finder - creator-content-calendar
skills/creator-content-calendar - crisis-comms-external
skills/crisis-comms-external - cross-platform-reformatter
skills/cross-platform-reformatter - csat-root-cause
skills/csat-root-cause - curriculum-mapper
skills/curriculum-mapper - customer-analytics
skills/customer-analytics - customer-success-qbr
skills/customer-success-qbr - dark-mode-design
skills/dark-mode-design - data-drift-monitor
skills/data-drift-monitor - data-quality
skills/data-quality - data-room-builder
skills/data-room-builder - data-story
skills/data-story - data-table-design
skills/data-table-design - database-schema
skills/database-schema - dead-code-eliminator
skills/dead-code-eliminator - debt-payoff-planner
skills/debt-payoff-planner - decision-making
skills/decision-making - deep-research
skills/deep-research - deep-work-planner
skills/deep-work-planner - demo-script
skills/demo-script - dependency-risk-audit
skills/dependency-risk-audit - design-critique
skills/design-critique - design-handoff-doc
skills/design-handoff-doc - design-qa-checklist
skills/design-qa-checklist - design-token-system
skills/design-token-system - differentiated-instruction
skills/differentiated-instruction - discovery-call-prep
skills/discovery-call-prep - docker-compose-wizard
skills/docker-compose-wizard - document-template-system
skills/document-template-system - donor-communications
skills/donor-communications - due-diligence-checklist
skills/due-diligence-checklist - e2e-scenario-author
skills/e2e-scenario-author - economic-analysis
skills/economic-analysis - eda-playbook
skills/eda-playbook - email-deliverability
skills/email-deliverability - email-drip-builder
skills/email-drip-builder - email-newsletter-pro
skills/email-newsletter-pro - email-triage
skills/email-triage - emergency-fund-planner
skills/emergency-fund-planner - employment-contract
skills/employment-contract - eng-status-rollup
skills/eng-status-rollup - engagement-reply-drafter
skills/engagement-reply-drafter - env-doctor
skills/env-doctor - error-budget-policy
skills/error-budget-policy - error-handling
skills/error-handling - error-message-writer
skills/error-message-writer - escalation-summary
skills/escalation-summary - executive-summary
skills/executive-summary - expansion-revenue
skills/expansion-revenue - expense-policy
skills/expense-policy - experiment-tracking
skills/experiment-tracking - expert-interview
skills/expert-interview - explain-plan-reader
skills/explain-plan-reader - fact-checker
skills/fact-checker - feature-flags
skills/feature-flags - feature-store-design
skills/feature-store-design - feedback-writer
skills/feedback-writer - fiction-scene-writer
skills/fiction-scene-writer - finance-fx
skills/finance-fx - financial-planner
skills/financial-planner - financial-statement-builder
skills/financial-statement-builder - fitness-program
skills/fitness-program - flaky-test-detangler
skills/flaky-test-detangler - flutter-widget-architect
skills/flutter-widget-architect - formatted-report-writer
skills/formatted-report-writer - fpa-model
skills/fpa-model - framework-upgrader
skills/framework-upgrader - freelance-positioning
skills/freelance-positioning - fun-content
skills/fun-content - fundraise-closing-mechanics
skills/fundraise-closing-mechanics - fundraise-readiness-audit
skills/fundraise-readiness-audit - fundraise-team-hiring
skills/fundraise-team-hiring - fundraising-narrative
skills/fundraising-narrative - fundraising-stage-selector
skills/fundraising-stage-selector - funnel-analysis
skills/funnel-analysis - geo-places
skills/geo-places - ghostwriter
skills/ghostwriter - git-commit-writer
skills/git-commit-writer - github-actions
skills/github-actions - go-to-market-planner
skills/go-to-market-planner - goals-accountability
skills/goals-accountability - google-business-profile-optimizer
skills/google-business-profile-optimizer - government-open-data
skills/government-open-data - grand-slam-offer-builder
skills/grand-slam-offer-builder - grant-loi-writer
skills/grant-loi-writer - grant-proposal-writer
skills/grant-proposal-writer - grant-prospect-researcher
skills/grant-prospect-researcher - graphql-schema
skills/graphql-schema - growth-accounting
skills/growth-accounting - growth-model
skills/growth-model - gtd-system
skills/gtd-system - gym-meta-ads-funnel
skills/gym-meta-ads-funnel - gym-money-model
skills/gym-money-model - gym-pricing-and-guarantees
skills/gym-pricing-and-guarantees - gym-transformation-challenge
skills/gym-transformation-challenge - habit-builder
skills/habit-builder - hashtag-strategist
skills/hashtag-strategist - help-documentation
skills/help-documentation - hiring-a-players-and-sops
skills/hiring-a-players-and-sops - hiring-pipeline
skills/hiring-pipeline - hiring-scorecard
skills/hiring-scorecard - icon-system
skills/icon-system - icp-persona-builder
skills/icp-persona-builder - idempotency-enforcer
skills/idempotency-enforcer - image-license-rights
skills/image-license-rights - impact-report-builder
skills/impact-report-builder - inbox-zero
skills/inbox-zero - index-advisor
skills/index-advisor - injury-prehab
skills/injury-prehab - instagram-post-builder
skills/instagram-post-builder - instagram-trend-scout
skills/instagram-trend-scout - internal-linking-mapper
skills/internal-linking-mapper - interview-debrief-synthesizer
skills/interview-debrief-synthesizer - interview-guide-builder
skills/interview-guide-builder - interview-question-kit
skills/interview-question-kit - interview-synthesis
skills/interview-synthesis - investment-basics
skills/investment-basics - investor-pipeline-crm
skills/investor-pipeline-crm - investor-targeting
skills/investor-targeting - investor-update-writer
skills/investor-update-writer - jd-bias-scrubber
skills/jd-bias-scrubber - jetpack-compose-builder
skills/jetpack-compose-builder - jira-ticket-writer
skills/jira-ticket-writer - job-application
skills/job-application - job-description-writer
skills/job-description-writer - job-interview-prep
skills/job-interview-prep - journal-framework
skills/journal-framework - jtbd-extractor
skills/jtbd-extractor - kafka-pipelines
skills/kafka-pipelines - kb-article-writer
skills/kb-article-writer - keyword-cluster-builder
skills/keyword-cluster-builder - kinetic-typography
skills/kinetic-typography - kpi-scoreboard-and-cadence
skills/kpi-scoreboard-and-cadence - kubernetes-basics
skills/kubernetes-basics - landing-page-copy
skills/landing-page-copy - landing-page-cro
skills/landing-page-cro - language-learning
skills/language-learning - language-reference
skills/language-reference - language-version-migrator
skills/language-version-migrator - launch-day-runbook
skills/launch-day-runbook - launch-email-sequence
skills/launch-email-sequence - launch-plan-sequencer
skills/launch-plan-sequencer - lead-enrichment
skills/lead-enrichment - lesson-plan-builder
skills/lesson-plan-builder - life-coach
skills/life-coach - lifecycle-journey-map
skills/lifecycle-journey-map - linear-workflow
skills/linear-workflow - linkedin-post-writer
skills/linkedin-post-writer - listing-description-writer
skills/listing-description-writer - literature-review
skills/literature-review - llm-evaluation
skills/llm-evaluation - load-testing
skills/load-testing - local-referral-engine
skills/local-referral-engine - local-seo-playbook
skills/local-seo-playbook - local-unit-economics
skills/local-unit-economics - localization-guide
skills/localization-guide - logo-brief-writer
skills/logo-brief-writer - longevity-protocol
skills/longevity-protocol - m-and-a-analysis
skills/m-and-a-analysis - manuscript-reviser
skills/manuscript-reviser - market-sizing
skills/market-sizing - marketing-attribution
skills/marketing-attribution - mcp-builder
skills/mcp-builder - meal-planner
skills/meal-planner - media-monitor
skills/media-monitor - meeting-agenda
skills/meeting-agenda - meeting-notes-to-actions
skills/meeting-notes-to-actions - messaging-hierarchy
skills/messaging-hierarchy - meta-title-optimizer
skills/meta-title-optimizer - microservices
skills/microservices - migration-safety-checker
skills/migration-safety-checker - ml-feature-engineering
skills/ml-feature-engineering - mobile-offline-sync
skills/mobile-offline-sync - mobile-perf-profiler
skills/mobile-perf-profiler - mobility-routine
skills/mobility-routine - mock-stub-designer
skills/mock-stub-designer - model-card-writer
skills/model-card-writer - model-evaluation-report
skills/model-evaluation-report - mongodb-expert
skills/mongodb-expert - monolith-decomposer
skills/monolith-decomposer - monorepo-expert
skills/monorepo-expert - month-end-close
skills/month-end-close - moodboard-builder
skills/moodboard-builder - motion-color-and-light
skills/motion-color-and-light - motion-design-principles
skills/motion-design-principles - motion-spec
skills/motion-spec - mutation-test-runner
skills/mutation-test-runner - mutual-action-plan
skills/mutual-action-plan - n-plus-one-hunter
skills/n-plus-one-hunter - narration-script
skills/narration-script - networking-system
skills/networking-system - next-on-vercel-perf
skills/next-on-vercel-perf - nextjs-app-router
skills/nextjs-app-router - no-show-reduction
skills/no-show-reduction - nonprofit-board-pack
skills/nonprofit-board-pack - notion-database
skills/notion-database - nutrition-planner
skills/nutrition-planner - oauth-flow
skills/oauth-flow - objection-handler
skills/objection-handler - objection-handling-and-speed-to-lead
skills/objection-handling-and-speed-to-lead - observability-stack
skills/observability-stack - offer-rejection-writer
skills/offer-rejection-writer - okr-builder
skills/okr-builder - onboarding-copy
skills/onboarding-copy - onboarding-plan-builder
skills/onboarding-plan-builder - oncall-handoff
skills/oncall-handoff - one-pager-designer
skills/one-pager-designer - op-ed-writer
skills/op-ed-writer - open-house-follow-up
skills/open-house-follow-up - outreach-sequence-designer
skills/outreach-sequence-designer - pagination-and-sync-engineer
skills/pagination-and-sync-engineer - paid-acquisition-audit
skills/paid-acquisition-audit - pandas-expert
skills/pandas-expert - partition-planner
skills/partition-planner - partnership-strategy
skills/partnership-strategy - patent-prior-art
skills/patent-prior-art - performance-review-writer
skills/performance-review-writer - pii-scrubber
skills/pii-scrubber - pitch-deck-builder
skills/pitch-deck-builder - pitch-practice-coach
skills/pitch-practice-coach - playwright-testing
skills/playwright-testing - plg-motion-designer
skills/plg-motion-designer - podcast-pitch
skills/podcast-pitch - policy-brief
skills/policy-brief - positioning-statement
skills/positioning-statement - postmortem-writer
skills/postmortem-writer - power-bi-dax
skills/power-bi-dax - pr-description-writer
skills/pr-description-writer - press-release-writer
skills/press-release-writer - pricing-strategy
skills/pricing-strategy - primary-research
skills/primary-research - print-layout
skills/print-layout - process-doc
skills/process-doc - product-analytics
skills/product-analytics - product-demo-director
skills/product-demo-director - product-description-writer
skills/product-description-writer - product-hunt-launch
skills/product-hunt-launch - product-roadmap
skills/product-roadmap - productized-service-designer
skills/productized-service-designer - prompt-engineer
skills/prompt-engineer - prompt-to-skill
skills/prompt-to-skill - prospect-list-builder
skills/prospect-list-builder - prospecting-metrics
skills/prospecting-metrics - prototype-spec
skills/prototype-spec - public-data-api-picker
skills/public-data-api-picker - push-notification-copy
skills/push-notification-copy - push-notification-wirer
skills/push-notification-wirer - python-async
skills/python-async - query-rewriter
skills/query-rewriter - quiz-generator
skills/quiz-generator - quote-follow-up-sequences
skills/quote-follow-up-sequences - r-for-analysis
skills/r-for-analysis - raise-your-rates-playbook
skills/raise-your-rates-playbook - rate-limit-handler
skills/rate-limit-handler - re-negotiation-prep
skills/re-negotiation-prep - react-native-pro
skills/react-native-pro - readme-generator
skills/readme-generator - redline-annotation
skills/redline-annotation - referral-and-affiliate-system
skills/referral-and-affiliate-system - refund-deescalation
skills/refund-deescalation - regex-master
skills/regex-master - regulatory-scan
skills/regulatory-scan - remotion-compose
skills/remotion-compose - remotion-render
skills/remotion-render - remotion-setup
skills/remotion-setup - replication-lag-debugger
skills/replication-lag-debugger - research-readout
skills/research-readout - research-synthesis
skills/research-synthesis - responsive-spec
skills/responsive-spec - resume-reviewer
skills/resume-reviewer - resume-writer
skills/resume-writer - retainer-economics-calculator
skills/retainer-economics-calculator - retainer-pricing-calculator
skills/retainer-pricing-calculator - retention-and-churn-killer
skills/retention-and-churn-killer - retirement-projection
skills/retirement-projection - revenue-modeling
skills/revenue-modeling - revenue-operations
skills/revenue-operations - review-engine
skills/review-engine - review-to-faq-builder
skills/review-to-faq-builder - rfp-response-writer
skills/rfp-response-writer - rubric-builder
skills/rubric-builder - runbook-writer
skills/runbook-writer - rust-patterns
skills/rust-patterns - saas-pricing
skills/saas-pricing - safe-vs-priced-round
skills/safe-vs-priced-round - salary-negotiation
skills/salary-negotiation - sales-enablement-kit
skills/sales-enablement-kit - sales-followup-cadence
skills/sales-followup-cadence - sales-proposal-writer
skills/sales-proposal-writer - scope-creep-defense
skills/scope-creep-defense - screencast-capture
skills/screencast-capture - screening-rubric-builder
skills/screening-rubric-builder - search-intent-classifier
skills/search-intent-classifier - second-brain
skills/second-brain - secrets-hygiene
skills/secrets-hygiene - secure-code-review
skills/secure-code-review - security-audit
skills/security-audit - segmentation-strategy
skills/segmentation-strategy - seller-lead-nurture
skills/seller-lead-nurture - seo-optimizer
skills/seo-optimizer - series-a-readiness
skills/series-a-readiness - sermon-writer
skills/sermon-writer - serp-gap-analyzer
skills/serp-gap-analyzer - sev-triage
skills/sev-triage - show-notes-writer
skills/show-notes-writer - side-project-launch
skills/side-project-launch - skill-auditor
skills/skill-auditor - skill-author
skills/skill-author - skill-creator
skills/skill-creator - skill-description-writer
skills/skill-description-writer - skill-pack-curator
skills/skill-pack-curator - skill-tester
skills/skill-tester - skillme
skills/skillme - sleep-optimizer
skills/sleep-optimizer - slide-deck-builder
skills/slide-deck-builder - soc2-evidence-helper
skills/soc2-evidence-helper - social-caption-writer
skills/social-caption-writer - social-content-calendar
skills/social-content-calendar - social-hook-generator
skills/social-hook-generator - social-video-formatter
skills/social-video-formatter - sound-and-music-sync
skills/sound-and-music-sync - space-earth-science
skills/space-earth-science - spark-jobs
skills/spark-jobs - speech-writer
skills/speech-writer - spreadsheet-model-builder
skills/spreadsheet-model-builder - sprint-planning
skills/sprint-planning - sprint-retro-facilitator
skills/sprint-retro-facilitator - sql-query-optimizer
skills/sql-query-optimizer - sql-to-insights
skills/sql-to-insights - stakeholder-update
skills/stakeholder-update - statement-of-work-writer
skills/statement-of-work-writer - status-page-update
skills/status-page-update - stock-photo-api
skills/stock-photo-api - stock-photo-finder
skills/stock-photo-finder - story-structure-architect
skills/story-structure-architect - strangler-fig-planner
skills/strangler-fig-planner - strength-training-plan
skills/strength-training-plan - stress-management
skills/stress-management - stripe-integration
skills/stripe-integration - student-feedback-writer
skills/student-feedback-writer - study-system
skills/study-system - supabase-expert
skills/supabase-expert - supply-chain-intel
skills/supply-chain-intel - support-macro-library
skills/support-macro-library - support-ticket-reply
skills/support-ticket-reply - survey-designer
skills/survey-designer - swift-ui
skills/swift-ui - systematic-review
skills/systematic-review - tableau-best-practices
skills/tableau-best-practices - tax-optimization
skills/tax-optimization - tdd-expert
skills/tdd-expert - team-charter
skills/team-charter - team-health-check
skills/team-health-check - tech-debt-prioritizer
skills/tech-debt-prioritizer - technical-blog-engine
skills/technical-blog-engine - technical-spec
skills/technical-spec - term-sheet-explainer
skills/term-sheet-explainer - term-sheet-negotiation
skills/term-sheet-negotiation - terms-of-service
skills/terms-of-service - terraform-expert
skills/terraform-expert - test-data-builder
skills/test-data-builder - test-placeholder-data
skills/test-placeholder-data - testimonial-capture-interview
skills/testimonial-capture-interview - threat-model-stride
skills/threat-model-stride - thumbnail-concept
skills/thumbnail-concept - time-series
skills/time-series - trend-analysis
skills/trend-analysis - trend-educator
skills/trend-educator - tweet-thread-builder
skills/tweet-thread-builder - typescript-strict
skills/typescript-strict - typography-system
skills/typography-system - unit-economics
skills/unit-economics - usability-test-plan
skills/usability-test-plan - user-flow-mapper
skills/user-flow-mapper - user-persona
skills/user-persona - ux-writing-audit
skills/ux-writing-audit - variant-copy-scaler
skills/variant-copy-scaler - vc-pitch-email
skills/vc-pitch-email - vercel-ai-gateway
skills/vercel-ai-gateway - vercel-deploy-pipeline
skills/vercel-deploy-pipeline - vercel-edge-and-isr
skills/vercel-edge-and-isr - vercel-env-management
skills/vercel-env-management - vercel-firewall-and-botid
skills/vercel-firewall-and-botid - video-hook-writer
skills/video-hook-writer - video-storyboard
skills/video-storyboard - visual-asset-curation
skills/visual-asset-curation - visual-hierarchy
skills/visual-hierarchy - vulnerability-triage
skills/vulnerability-triage - weather-climate
skills/weather-climate - web-performance
skills/web-performance - webhook-receiver-hardener
skills/webhook-receiver-hardener - webinar-funnel-builder
skills/webinar-funnel-builder - websocket-expert
skills/websocket-expert - weekly-review
skills/weekly-review - white-space-analysis
skills/white-space-analysis - whitepapers
skills/whitepapers - wikipedia-style
skills/wikipedia-style - win-back-campaign
skills/win-back-campaign - youtube-script-writer
skills/youtube-script-writer - zone-2-cardio-plan
skills/zone-2-cardio-plan
How it works
1on1-agenda
When your agent uses it
Structures recurring manager-report 1:1s with a rolling shared agenda that balances check-in, blockers, two-way feedback, and career growth instead of status updates. Use when someone asks "what should I cover in my 1:1s", "build me a 1:1 agenda template", "my 1:1s have turned into status meetings", or is setting up 1:1s with new reports. Do NOT use for general meeting agendas - use meeting-agenda instead; for writing a formal review document, use performance-review-writer; for drafting a specific piece of feedback, use feedback-writer.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,475 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core principle
- Operating procedure
- Template: rolling 1:1 doc
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,394 bytes
Read SKILL.md
# 1:1 Agenda
The 1:1 is the manager's highest-leverage meeting, and it belongs to the report. The costly failure it prevents is invisible until it isn't: the resignation that "came out of nowhere", the burnout nobody asked about, the career stagnation that festered because growth only came up at review season. A structured rolling agenda keeps the meeting about the human, not the task list.
## Core principle
The 1:1 is their meeting. Status belongs in standups and written updates; this time is protected for what does not fit there - blockers, growth, feedback, and how they are actually doing. The report should lead roughly 70% of the talking. If the manager is doing most of it, the meeting has become a broadcast.
## Operating procedure
### Step 1: Gather inputs
Before setting up (or fixing) a 1:1 practice, establish:
ab-test-analyzer
When your agent uses it
Analyzes an A/B experiment to a defensible verdict - sample-size and minimum-detectable-effect math, two-proportion significance testing with confidence intervals, and checks for the traps that fake a win (peeking, sample ratio mismatch, multiple comparisons). Use when someone asks "is this test significant", "did variant B win", "how long should I run this experiment", "what sample size do I need", or shows experiment results and asks whether to ship. Do NOT use for logging and organizing ML training runs - use experiment-tracking instead; for causal questions without a randomized experiment use causal-inference; for general funnel or metric investigation use funnel-analysis or product-analytics.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
2,364 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: readout with the correct interpretation
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore9,394 bytesSKILL.mdcore9,394 bytes
Read SKILL.md
# A/B Test Analyzer
Read experiments like a skeptic: most "wins" are noise until proven otherwise. The costly mistake this skill prevents is shipping on a peeked, underpowered, or mis-split test - a false positive that ships is worse than no test, because the team now defends a change that does nothing. Every verdict here comes with the interval, the checks that passed, and the checks that did not.
## Operating procedure
Order is mandatory: validity checks (Step 3) come before the significance math (Step 4), because a significant result on an invalid test is still invalid.
### Step 1: Gather inputs
Collect before computing anything; label unknowns as guesses.
1. The single primary metric and its type (conversion rate, revenue per user, retention).
2. The hypothesis and the minimum effect that would matter to the business - the smallest lift worth the cost of shipping and maintaining the change. If the requester has no number, help set one; without it "significant" is uninterpretable.abandoned-cart-sequence
When your agent uses it
Drafts a margin-aware 3-email cart-recovery sequence that reminds, handles objections, then incentivizes only as a last touch. Use when a checkout is started but not completed and you need cart-recovery (abandoned-cart) emails, when wiring an abandoned-cart flow in Shopify, Klaviyo, or similar, or when asked to recover lost carts without overspending on discounts.
What it loads into context
Frontmatter
98 tokensAlways available to the agent
SKILL.md
757 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
Files
SKILL.mdcore3,446 bytes
Read SKILL.md
# Abandoned Cart Sequence
Produce a three-email cart-recovery arc that recovers the sale while protecting margin: remind, then reassure, then incentivize only if needed.
## Workflow
1. Gather inputs before drafting: brand voice, average order value, margin floor (the discount level below which a sale loses money), whether free shipping is available, and the top one or two abandonment objections. If any are unknown, ask rather than guess.
2. Draft Email 1 - Helpful Reminder (send 1-4 hours after abandon, while intent is warm). Helpful tone, not salesy. Show cart contents with image and name, restate the core benefit in one line, one clear button back to the cart. No discount. Subject line names the product or gives a friendly nudge, not a generic "Did you forget something?".
3. Draft Email 2 - Objection Handler (send ~24 hours later). Surface trust signals (reviews, ratings, guarantee, easy returns, secure checkout), answer the top objection, add honest social proof ("4.8 stars from 2,000 buyers"). Still no discount - at most a free-shipping nudge if margins allow, since shipping cost is the most common abandonment cause. Use urgency claims ("items in your cart are popular") only when literally true.
4. Draft Email 3 - Incentive + Honest Urgency (send ~48-72 hours later). If and only if margin allows, offer one incentive framed as a thank-you, not a fire sale. State the code clearly and link to a pre-filled cart. This is the final touch; do not repeat the same incentive.
5. Apply incentive logic: discount last, never first. Prefer free shipping or a small gift over deep percentage cuts on low-margin items; use a fixed-dollar amount for high-AOV carts; never go below the margin floor.
6. Push back when the brief is self-defeating: if the brand wants a discount in Email 1, flag that it erodes margin and trains customers to abandon for a coupon; if AOV is low and margins thin, recommend a free-shipping threshold or no incentive instead of a percentage off.
## Quality bar
- Exactly three emails, each with a stated send delay, a distinct job (remind / reassure / incentivize), and a single primary CTA back to the cart.
- No discount appears before Email 3, and no incentive ever breaches the stated margin floor.academic-essay
When your agent uses it
Structures and drafts argumentative academic essays - a contestable thesis, claim-evidence-analysis body paragraphs, a fairly-engaged counterargument, and a synthesizing conclusion - marking every spot that needs a real citation instead of inventing sources. Use when a student or writer asks "help me outline my essay", "is my thesis strong enough", "structure my argument for this paper", or needs a draft refined for logic and formal register. Do NOT use for opinion pieces aimed at a general audience - use op-ed-writer instead; for surveying scholarly sources, use literature-review.
What it loads into context
Frontmatter
131 tokensAlways available to the agent
SKILL.md
828 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Integrity note
- Process
- Anatomy of the essay
- Writing rules
- Strengthening the argument
- Anti-patterns
- Output
Files
SKILL.mdcore3,972 bytesSKILL.mdcore3,972 bytes
Read SKILL.md
# Academic Essay
You help structure and write argumentative academic essays. An academic essay advances a contestable thesis and defends it with evidence and reasoning, engaging seriously with opposing views.
## Integrity note
Support the writer's thinking - outline, structure, strengthen argument, improve clarity. Encourage the writer to do their own analysis and cite genuinely. Don't fabricate sources, quotes, or data; a fake citation is academic misconduct. Flag where real sources must be supplied.
## Process
1. Clarify the prompt/question and the discipline's conventions (history, literature, science differ).
2. Develop a clear, arguable thesis.
3. Outline the argument, then draft, then refine for logic and flow.
accessibility-audit
When your agent uses it
Runs a full WCAG 2.2 accessibility audit of an interface - keyboard, screen reader, contrast and zoom, forms, and media, in a fixed pass order - and delivers a severity-triaged findings table with concrete fixes. Use when someone asks "is this page accessible", "run an a11y audit on this UI", "will this pass WCAG AA", "why can't a screen reader use this form", or is preparing for an accessibility review, procurement questionnaire, or VPAT. Do NOT use for color-and-contrast-only questions such as palette checks or colorblind-safe data visualization - use color-accessibility instead.
What it loads into context
Frontmatter
146 tokensAlways available to the agent
SKILL.md
1,803 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Findings table template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,834 bytes
Read SKILL.md
# Accessibility Audit
An interface that fails keyboard or screen reader users fails them completely - not "slightly worse UX," but locked out. Retrofitting accessibility after launch costs several times what building it in does, and automated scanners catch only about 30% of real defects, so a team that "ran axe and passed" usually ships the other 70%. This audit finds the full set, in an order that stops early failures from masking later ones, and triages every finding by severity so the team fixes blockers before polish.
## Inputs to collect
Gather these before auditing. If any is missing, ask; if the user shrugs, apply the default and label it a guess.
1. **Scope** - the pages, flows, or components under audit. Default: the top task flow (e.g., sign-up through first core action), not the whole site.
2. **Target conformance level** - default WCAG 2.2 Level AA. Level A alone is rarely acceptable for procurement; AAA is aspirational, not a default.
3. **Artifact type** - live URL, code, or screenshots. Code beats screenshots: focus order and ARIA state are invisible in a picture.
4. **Assistive tech matrix** - default VoiceOver + Safari (macOS/iOS) and NVDA + Chrome (Windows).
5. **Known constraints** - legacy widgets, third-party embeds, legal deadlines.
ad-creative-testing
When your agent uses it
Designs a disciplined ad creative testing system with hooks, isolated variables, and statistical rigor. Use when setting up a testing pipeline or when creative performance is declining.
What it loads into context
Frontmatter
50 tokensAlways available to the agent
SKILL.md
1,220 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Structure Tests Around a Single Variable
- Prioritize Hook Testing First
- Set Sample Sizes Before Running
- Maintain a Creative Log
- Iteration Rhythm
- Worked example
- Deliverable
- Common Mistakes to Avoid
- Quality bar
Files
SKILL.mdcore5,841 bytes
Read SKILL.md
# Ad Creative Testing
Most ad creative testing produces noise, not learning. Discipline means changing one variable at a time, running tests long enough, and acting only on statistically meaningful results.
## Structure Tests Around a Single Variable
Each test must isolate exactly one variable: hook (first 3 seconds of video or headline), visual format (static vs. video vs. carousel), offer framing (discount vs. free trial vs. guarantee), or call to action. Testing multiple variables at once makes the result uninterpretable. Label every creative with the variable it is testing and the hypothesis it is checking before launch.
## Prioritize Hook Testing First
The hook is the highest-leverage variable in most paid channels. For video, the 3-second view rate and thumb-stop ratio reveal hook strength before the full watch occurs. For static and search, the headline carries the same weight. Run hook tests before any other variable - a great offer with a weak hook will never be seen. A hook passes if 3-second view rate exceeds 30% on Meta or thumb-stop exceeds 25%; a headline passes if CTR exceeds channel benchmark.
## Set Sample Sizes Before Running
aeo-answer-blockifier
When your agent uses it
Rewrites a section or page into concise, self-contained answer blocks that AI answer engines and search snippets can lift and cite verbatim. Use when restructuring a blog post, FAQ, docs page, or landing copy to win a featured snippet or get quoted by AI Overviews, ChatGPT, or Perplexity, or when a draft buries its answer in setup. Do NOT use for keyword research, full-article drafting, or technical/crawlability SEO - use a dedicated SEO or copywriting skill instead.
What it loads into context
Frontmatter
125 tokensAlways available to the agent
SKILL.md
906 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,164 bytes
Read SKILL.md
# AEO Answer Blockifier
Restructure existing content into question-led blocks an engine can quote out of context, so the page earns citations and snippets instead of just ranking.
## Workflow
1. **Extract the questions.** List every distinct question the section actually answers, in the user's natural wording (including conversational and voice phrasing). One block per question. If a passage answers no clear question, either reframe it as one or drop it from blockification.
2. **Set the heading to the question.** Make each H2/H3 the exact question, matching how a user would type or speak it. This maps the block to the query and mirrors People Also Ask.
3. **Lead with the answer.** State the direct answer in the first one or two sentences under the heading - conclusion first, before any setup. Supporting detail, nuance, and caveats follow. An answer buried in paragraph three forfeits the citation; extraction overwhelmingly favors the first 1-2 sentences after the heading.
4. **Make the block self-contained.** Name the subject explicitly in the answer sentence. Remove "as mentioned above," "this," and any pronoun or reference pointing offscreen, so the block reads correctly when lifted alone.
5. **Pick the format the question demands.** "How to" / "steps" → numbered list. "Best" / "types of" → bulleted list. "What is" → one tight definition paragraph. Comparison → a table. Forcing prose onto a list-shaped query loses the extraction.
6. **Size the block to the format.** Definition/paragraph answer: roughly 40-60 words (featured snippets typically truncate near 300 characters, so the complete answer must land before that cutoff). List: 3-8 scannable items, one idea each, under ~10 words per item where possible. Table: labeled columns, no prose padding in cells.
7. **Add verifiable trust signals.** Where the answer makes a factual claim, attach a concrete number, named source, or first-hand specific an engine can check (this is what E-E-A-T rewards). Replace generic unsourced assertions with verifiable ones.
8. **Verify against the quality bar** before returning.
agency-monthly-report
When your agent uses it
Build the monthly client report that renews retainers - results-first structure with business outcomes before activity, the metric-that-matters-to-them rule, work-completed as evidence rather than headline, and a next-month plan with one client-action recommendation. Includes a full copy-paste template and a good/bad results-section pair. Use when an agency team says "write this month's client report", "our reports are just a list of tasks we did", "the client says they don't know what they're paying for", or "make our reporting actually defend the retainer". Do NOT use for the quarterly business review deck - use agency-qbr-upsell instead.
What it loads into context
Frontmatter
142 tokensAlways available to the agent
SKILL.md
1,841 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: good/bad results-section pair
- Template: monthly report
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore8,031 bytes
Read SKILL.md
# Agency Monthly Report
The monthly report is the one artifact every stakeholder on the client side sees - including the budget holder who never joins calls - and it is where retainers are silently renewed or silently cancelled. The costly mistake this skill prevents is the activity report: a proud list of tasks completed that forces the client to do the math on whether any of it mattered. Clients do not renew for effort; they renew because the report proves the retainer moves a number they care about.
Work the example agency throughout: Northbeam Digital, an 8-person marketing agency with 11 retainer clients averaging $6,500/month and a 62 percent gross margin target. The report is the monthly rung of the cadence ladder in client-comms-cadence, delivered on the fifth business day every month.
## Operating procedure
### Step 1: Fix the metric that matters to them
Before writing anything, confirm the one business metric this client's budget holder is judged on - revenue, qualified leads, cost per acquisition, booked calls. Not the agency's favorite metric; theirs. It was set at onboarding (client-onboarding-system) and re-confirmed at each QBR. If it has never been named, that is the first agenda item of the next call, and this month's report should say so honestly rather than guess. Every report leads with this metric. If the agency cannot connect its work to it, the retainer is already in danger regardless of how good the report looks.
### Step 2: Write the results section first
agency-positioning
When your agent uses it
Reposition a generalist agency around a vertical or service specialization, write the "we're the agency that X for Y" statement, and design a paid-audit entry offer backed by aligned case studies. Use when an agency owner says "we do everything for everyone and it's killing our pricing", "should we niche down", "prospects can't tell us apart from every other agency", or "how do we productize the front of our funnel". Do NOT use for writing a one-line positioning statement for a product or startup - use positioning-statement instead.
What it loads into context
Frontmatter
120 tokensAlways available to the agent
SKILL.md
1,520 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore7,164 bytes
Read SKILL.md
# Agency Positioning
A generalist agency competes on price against every other generalist, wins only by referral, and builds case studies that never compound. Specialization is what lets an agency charge for outcomes instead of hours, and the costly mistake this skill prevents is the fake niche: a website that claims a specialty the client list, the case studies, and the entry offer do not back up. Positioning is only real when all four line up.
Work the example agency throughout: Northbeam Digital, an 8-person marketing agency with 11 retainer clients averaging $6,500/month ($71.5k MRR) and a 62 percent gross margin target.
## Operating procedure
Follow the steps in order. The client-base audit comes first because the right specialization is almost always already hiding in the current book of business - you pick the niche you can prove, not the one that sounds best.
### Step 1: Audit the current client base
Pull per-client margin and revenue from retainer-economics-calculator (install that skill first - its margin numbers drive this decision). For each of the clients, record: vertical, primary service consumed, gross margin, retainer size, and whether the team likes the work. If margin data is a guess, label it a guess.
agency-qbr-upsell
When your agent uses it
Run agency quarterly business reviews that expand accounts - the four-part QBR agenda (results retrospective, goal re-alignment, roadmap, expansion proposal), an expansion trigger inventory, land-and-expand sequencing, and price-increase timing. Use when an agency owner says "our QBRs are just bigger status meetings", "how do I grow existing retainers", "when do I raise this client's price", or "the client's goals changed and our scope didn't". Do NOT use for SaaS customer-success QBRs run by a CS team - use customer-success-qbr instead.
What it loads into context
Frontmatter
125 tokensAlways available to the agent
SKILL.md
1,770 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: expansion proposal skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore7,897 bytesSKILL.mdcore7,897 bytes
Read SKILL.md
# Agency QBR Upsell
The cheapest revenue an agency will ever win is the next $2,000/month from a client already paying $6,500 - no pitch, no audit, no onboarding cost, and it lands at existing margin. The costly mistake this skill prevents is the QBR-as-status-meeting: ninety minutes of recap with no proposal, which wastes the one scheduled moment each quarter when the client's decision-makers are in the room expecting to talk about the future. Every QBR ends with a forward-looking ask - an expansion, a renewal confirmation, or a repriced scope.
Work the example agency throughout: Northbeam Digital, an 8-person marketing agency with 11 retainer clients averaging $6,500/month and a 62 percent gross margin target. The QBR is the top rung of the cadence ladder in client-comms-cadence; a founder attends every one.
## Operating procedure
### Step 1: Prepare from the record, and gate on capacity
Pull the quarter's three monthly reports from agency-monthly-report - the QBR retrospective is those reports rolled up, never a new analysis the client has not seen. Pull the client's margin from retainer-economics-calculator, because margin decides the play:
- Client at or above the 50 percent floor: expansion candidate.
- Client below 50 percent: the QBR proposal is a reprice or re-scope, not more scope at the same broken rate.agent-orchestration
When your agent uses it
Designs reliable multi-agent LLM systems - choosing a topology, writing handoff contracts between agents, deciding what runs in parallel vs series, and setting context, retry, and cost budgets that stop runaway loops. Use when someone asks "should I split this into multiple agents", "my agents keep looping", "how do I pass context between agents", "orchestrator vs pipeline vs router", or is architecting an agent workflow. Do NOT use for tuning a single prompt - use prompt-engineer instead; do NOT use for measuring agent output quality - use llm-evaluation instead; for making a product agent-operable end to end, use build-on-agent-native.
What it loads into context
Frontmatter
154 tokensAlways available to the agent
SKILL.md
1,771 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: handoff contract
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,115 bytes
Read SKILL.md
# Agent Orchestration
Multi-agent systems fail two ways: teams split work across agents that a single capable agent handles better (paying latency, cost, and failure surface for nothing), or they wire agents together with vague handoffs and no budgets, producing loops that burn tokens until someone notices the bill. This skill designs the coordination layer - topology, contracts, budgets - so agent count is justified and every failure mode has a pre-decided response.
## Operating procedure
### Step 1: Gather inputs
- The end-to-end task and the acceptance criteria for its final output.
- Why one agent is insufficient - name the specific failure (context overflow, skill mismatch, needed parallelism). If the user cannot name one, the answer is one agent, and this design is over.
- Volume of context the task consumes vs the model's window; latency and cost ceilings per run.
- Which steps have side effects (writes, sends, purchases) vs are pure reads.
### Step 2: Justify each agentalert-tuning
When your agent uses it
Reduces alert noise by making alerts actionable, symptom-based, and tied to SLOs. Identifies and kills flappy, duplicate, and cause-based alerts. Use when alert fatigue is degrading on-call response quality.
What it loads into context
Frontmatter
62 tokensAlways available to the agent
SKILL.md
704 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The Audit - Run First
- Symptom vs Cause
- SLO-Based Alerting
- Fixing Flappy Alerts
- Routing and Severity
- Do / Don't
Files
SKILL.mdcore3,102 bytes
Read SKILL.md
# Alert Tuning
Every alert that fires without requiring human action trains engineers to ignore alerts. Alert fatigue kills real incidents. The goal: every page wakes someone up for a reason they cannot automate away.
## The Audit - Run First
Before writing new alert rules, audit what you have.
1. Pull the last 30 days of alert history. Calculate: total pages, pages outside business hours, pages that resulted in no action taken.
2. Identify flappy alerts: any alert that fires and resolves more than 3 times in a 24-hour window without a corresponding incident.
3. Identify cause-based alerts: anything that fires on a system-internal metric (queue depth, pod restart count, CPU) without a direct user-impact correlation.
4. Calculate the actionability rate: (pages with a documented remediation / total pages). Below 80% means your alert set is broken.
## Symptom vs Causeamazon-listing-optimizer
When your agent uses it
Writes an Amazon product title, five bullets, and backend search terms that obey Amazon's character, byte, and keyword rules while staying click-worthy in the search grid. Use when drafting or fixing a Seller Central listing, rewriting an Amazon title or bullet points, packing backend search terms, or recovering a listing flagged or suppressed for content. Do NOT use when writing on-site product-detail-page copy for your own storefront or marketing site - use product-description-writer instead.
What it loads into context
Frontmatter
113 tokensAlways available to the agent
SKILL.md
826 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore3,890 bytes
Read SKILL.md
# Amazon Listing Optimizer
Optimize an Amazon listing for both the A9 search algorithm and a shopper scanning a crowded grid, within Amazon's exact structural rules.
## Workflow
1. Gather inputs: the product, top 5-10 target keywords, brand name, key specs, and the category. If keywords are missing, ask for them or for the category to infer intent - do not guess. If a requested claim violates Amazon policy or lacks substantiation, refuse and propose compliant phrasing before writing.
2. Write the title. Cap under 200 characters, but many categories cap at 80 - target roughly 80-150 and verify the category limit. Order: brand, primary keyword, then most search-relevant attributes (type, size, quantity, color). Capitalize the first letter of each major word. Make the first ~50 characters say what it is and why it differs, since that is what shows on mobile.
3. Write exactly 5 bullets. Begin each with a 2-4 word capitalized benefit label, then one supporting sentence. Keep each bullet under ~200 characters for mobile readability. Cover in priority order: main benefit, key feature, fit/compatibility, what is in the box or dimensions, trust/guarantee point. Reserve one bullet for the top objection that reviews reveal.
4. Write backend search terms. The field caps at 250 bytes total (not per line); spaces separate terms, so no commas. Do not repeat any word already in the title or bullets - Amazon indexes those separately and repeats waste bytes. Add synonyms, common and alternate spellings, and use-case terms. Lowercase throughout.
5. Run the compliance pass before delivering: scan every field against the Do NOT list and the category's restricted-word rules (supplements, beauty, baby are strictest).
## Deliverable
Produce a paste-ready listing package containing:
animation-system
When your agent uses it
Builds a coherent UI motion system - a duration token scale, named easing curves, choreography rules, and a per-component mapping - so every transition in the product draws from one vocabulary instead of ad-hoc values. Use when someone asks "define our motion tokens", "our animations feel inconsistent", "what duration should this modal use", "set up easing standards", or is starting a design system and needs the motion layer. Do NOT use to spec the motion of one specific component or flow - use motion-spec instead; for animated type treatments, use kinetic-typography.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
2,011 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked token set (copy-paste artifact)
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,416 bytesSKILL.mdcore8,416 bytes
Read SKILL.md
# Animation System
Without a motion system, every engineer picks a duration and curve from memory, and the product accumulates dozens of near-identical transitions that read as sloppy even when no single one is wrong. This skill produces the token set and rules that make motion feel like one hand designed it - and prevents the expensive alternative: auditing and re-timing every transition after launch.
Motion must do a job: direct attention, show cause and effect, express hierarchy, or soften a change. A motion that does none of these gets cut - it is latency wearing a costume.
## Operating procedure
Durations come before easings, and both come before choreography, because choreography rules reference the tokens.
### Step 1: Gather inputs
1. Brand feel on a calm↔energetic axis - it shifts the scale ±50ms and decides whether springs are allowed.
2. Platforms (web, iOS, Android) and the animation stack (CSS, Framer Motion, native).annual-plan
When your agent uses it
Runs an annual planning cycle end to end - strategic reflection, 3-5 company priorities with explicit not-doing decisions, cascaded OKRs reconciled between top-down framing and bottom-up team plans, resource allocation with named trade-offs, a quarterly review cadence, and the CEO narrative memo. Use when a leader asks "run our annual planning", "set next year's company priorities", "our planning is just a budget exercise", or needs to turn a leadership offsite into a committed plan. Do NOT use for writing a single team's quarterly OKRs - use okr-builder instead; for a product-specific roadmap, use product-roadmap.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
830 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Start With Strategy, Not the Calendar
- The Planning Cadence
- Setting OKRs
- Resource Allocation
- The CEO Memo
- Keeping It Alive
- Anti-Patterns
- Deliverable
Files
SKILL.mdcore3,731 bytes
Read SKILL.md
# Annual Planning
Annual planning fails when it becomes a budget exercise disconnected from
strategy, or a wish-list with no prioritization. Done well, it produces a small
set of priorities the whole company can rally behind. This skill structures the
cycle end to end.
## Start With Strategy, Not the Calendar
Before any goals, answer the strategic questions:
- Where are we now, honestly? (Reflect on last year: what worked, what did not,
what surprised us.)
- What is the single most important thing we must achieve this year to win?api-client-generator
When your agent uses it
Generates a typed API client from an OpenAPI/Swagger spec, with a hand-controlled transport wrapper for timeouts, auth, and typed errors. Use when integrating a REST API that ships an openapi.yaml/swagger.json, when generating or regenerating a client from a spec, or when a hand-written client keeps drifting from the upstream contract. Do NOT use when the task is the backoff/retry policy itself - use rate-limit-handler instead; do NOT use for cursor pagination or keeping a local copy in sync - use pagination-and-sync-engineer instead.
What it loads into context
Frontmatter
129 tokensAlways available to the agent
SKILL.md
1,269 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Worked example: the transport seam
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,850 bytes
Read SKILL.md
# API Client Generator
Generate the API surface from the spec so types stay honest, then wrap the generated code in a transport layer you own for resilience.
## Workflow
1. Confirm a spec exists. If the API ships an OpenAPI/Swagger document, generate from it. If there is no spec, or the integration is a single endpoint, hand-write a small typed wrapper with the same transport in step 4 - skip codegen.
2. Generate the client. Run a codegen tool (openapi-typescript-codegen, openapi-generator, oapi-codegen, or equivalent) to produce request methods plus request/response types. Pin the spec by version or hash so an upstream change surfaces as a reviewable diff, not a silent runtime surprise.
3. Treat generated code as a build artifact. Commit it, regenerate it from the pinned spec, and never hand-edit it - edits get clobbered on the next regen.
4. Wrap, don't modify. Inject a custom HTTP transport (the client's configurable fetch/http layer) instead of editing generated files. Put timeouts, auth, default headers, and logging in that one wrapper so every generated call inherits them and the generated layer stays regenerable.
5. Set an explicit per-request timeout in the transport - generated clients usually default to none.
6. Make errors typed. Map non-2xx responses to a discriminated error type - auth (401/403), not-found (404), validation (422), rate-limit (429), server (5xx) - and parse the API's structured error body into a typed shape so callers branch on a tag, not on parsed strings. Never let a non-2xx return a half-empty success object.
7. Centralize config. Hold base URL, auth token refresh, and default headers in one configured instance; don't scatter raw fetch calls across the codebase.
8. Optionally validate critical responses against the schema at runtime so spec drift fails loudly. Generated types describe the contract, not reality.
9. Defer cross-cutting policy. Add the retry/backoff hook point in the transport, but get the backoff policy (exponential backoff, full jitter, Retry-After, retriable status codes, idempotency) from rate-limit-handler. Get cursor pagination and incremental sync from pagination-and-sync-engineer. This skill owns the generated surface and the transport seam, not those policies.
api-design
When your agent uses it
Designs REST API surfaces - resource naming, HTTP method and status-code semantics, error shapes, pagination, and filtering - and delivers an endpoint spec a consumer can build against without asking questions. Use when someone asks "how should I name this endpoint", "what status code should this return", "should this be PUT or PATCH", "how do I paginate this list", or is reviewing an API before it ships to external consumers. Do NOT use for planning breaking-change rollouts and deprecation windows - use api-versioning-strategist instead; for GraphQL type and resolver design - use graphql-schema instead; for generating client SDKs from an existing spec - use api-client-generator instead; for designing inbound webhook endpoints - use webhook-receiver-hardener instead.
What it loads into context
Frontmatter
168 tokensAlways available to the agent
SKILL.md
1,546 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,895 bytes
Read SKILL.md
# REST API Design
APIs are products: the consumer's mental model, not your database schema, decides the shape. The costly mistake this skill prevents is shipping an inconsistent surface - mixed naming, wrong status codes, offset pagination on a growing table - that every consumer then hardcodes around, making it effectively unfixable once integrated.
## Operating procedure
Work in this order. Resources come before methods because method semantics depend on what the resource is; error and pagination contracts come before the spec is written because retrofitting them breaks consumers.
### Step 1: gather inputs
Collect before designing. Label any assumption as a guess.
1. The consumers: internal services, first-party clients, or external developers. External consumers raise the consistency bar - default to external if unknown.
2. The core nouns of the domain and which ones own which (drives nesting).api-versioning-strategist
When your agent uses it
Produces an API version scheme (date-pinned header or URI), a breaking-vs-additive change policy, and a published deprecation/sunset timeline with translation shims. Use when removing or renaming a field or endpoint, tightening validation, cutting a "v2", binding a partner integration to a version, or planning how long an old version lives. Do NOT use when designing the resource shape, URLs, status codes, or pagination of a new endpoint - use REST API Design instead; this skill owns only the version scheme and deprecation path layered on top of that contract.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,126 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,398 bytes
Read SKILL.md
# API Versioning Strategist
Pick a version scheme, classify every change as breaking or additive, and ship a deprecation path that never silently breaks a live integration.
## Workflow
1. **Classify the change first.** Decide whether it is breaking or additive before choosing any mechanism.
- Breaking: removing/renaming a field or endpoint; adding a required request field; tightening validation; changing a field's type or an enum's semantics; changing error codes or status codes clients branch on.
- Additive (ship freely within the current version): adding optional request fields; adding response fields; adding endpoints; adding optional enum values clients are told to tolerate.
- Mandate a tolerant-reader contract: clients MUST ignore unknown fields. Document this so additive changes stay non-breaking.
2. **Choose the scheme to match release cadence.**
- Frequent, granular evolution → date-pinned versions: clients send a version header (a pinned release identifier), the account is bound to a default, and each release is a small documented set of breaking changes - ideally 1-5 per release, never dozens. Scales without parallel rewrites.
- Coarse, infrequent revisions → URI versioning (`/v1/`): simplest and most visible, but biases toward big-bang jumps. Pick only when breaking releases happen roughly once a year or less.
- Avoid letting a "v2" become a years-long parallel fork of "v1".
3. **Translate at the edge, never fork logic.** Keep one current internal model. Apply ordered version-transform shims that up-convert old requests and down-convert responses to a pinned client's contract. Each shim is small, independently testable, and removable when its version sunsets. Forking business logic per version is forbidden.apollo-prospecting
When your agent uses it
Use when executing your B2B prospecting strategy inside Apollo.io specifically - turning an ICP into stacked Apollo search filters, building and tiering Apollo Lists, finding and verifying emails with credit discipline, tracking buying signals via filters and saved-search alerts, and running multi-step Apollo Sequences. Trigger on "Apollo", "Apollo.io", "set up Apollo filters", "Apollo saved search", "Apollo list", "Apollo sequence", "find emails in Apollo", "Apollo credits", "export from Apollo to my CRM", "Apollo intent data", "how do I prospect in Apollo". Do NOT use for designing the ICP (use [[icp-persona-builder]]), writing email copy (use [[cold-email-craft]]), SPF/DKIM/DMARC deliverability (use [[cold-email-deliverability]]), or reading funnel metrics (use [[prospecting-metrics]]).
What it loads into context
Frontmatter
189 tokensAlways available to the agent
SKILL.md
3,148 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The workflow
- Apollo filter recipe (ICP → stacked filters)
- Apollo Sequence outline (cadence skeleton)
- Worked example (end to end, one ICP)
- Deliverable
- Quality bar
- Common failure modes
Files
SKILL.mdcore14,226 bytes
Read SKILL.md
# Prospect in Apollo.io
This is the tool-specific execution layer for the **B2B Prospecting Engine** pack. The methodology skills decide *what* to do; this skill shows you how to do it *in Apollo.io*. It assumes you have already built your ICP and personas in [[icp-persona-builder]], decided your list strategy in [[prospect-list-builder]], chosen your enrichment and verification rules in [[lead-enrichment]], picked the signals you care about in [[buying-signal-tracker]], structured your cadence in [[outreach-sequence-designer]], hardened your sending infrastructure in [[cold-email-deliverability]], and defined your funnel math in [[prospecting-metrics]].
If you have not done that work, stop and do it first. Apollo will happily let you blast a poorly-targeted list from your primary domain and torch both your deliverability and your data budget. The tool is a power amplifier, not a strategy.
Apollo changes its UI often, so this skill stays conceptual on *where* things live and concrete on *what to do and why*. When a specific menu or label is uncertain, confirm it in-product rather than trusting a click path from memory.
## When to use this skill
- You have an approved ICP and you need to translate it into Apollo People/Company Search filters.
- You are building or tiering lists in Apollo and need to suppress existing customers and open opportunities.
- You need to find and verify emails in Apollo without burning credits on garbage.
- You want Apollo to surface buying signals (job changes, hiring, funding, intent) and alert you.app-store-copy
When your agent uses it
Writes App Store and Play Store listing copy to platform-specific rules - iOS name, subtitle, and hidden keyword field; Android title, short description, and keyword-woven full description - delivering every field with character counts, a ranked keyword list, and title options to A/B test. Use when someone asks "write our app store description", "optimize our ASO keywords", "why isn't our app ranking in the App Store", or is preparing a listing for launch. Do NOT use for optimizing Amazon product listings - use amazon-listing-optimizer instead; for release notes and submission prep, use app-store-release-prep.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
818 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- iOS fields and limits
- Android (Play Store) fields and limits
- Description structure (both platforms)
- Keyword strategy
- Conversion rules
- Anti-patterns
- Output
Files
SKILL.mdcore3,653 bytes
Read SKILL.md
# App Store Copy
You write store listing copy that ranks in App Store / Play Store search AND converts browsers into installs. ASO and conversion are two jobs; you do both.
## Process
1. Gather: app name, what it does, top 3 features users love, target user, and key competitor apps.
2. Research the keyword vocabulary your users actually search.
3. Write each field to its platform's rules and limits.
## iOS fields and limits
- **App name** (30 chars): brand + a high-value keyword if room. "Notion - Notes & Docs."
- **Subtitle** (30 chars): your value prop, keyword-rich. Indexed for search.app-store-release-prep
When your agent uses it
Produce a reproducible signing, versioning, and store-declaration pipeline so an iOS or Android build passes App Store Connect / Play Console submission. Use when building or uploading a distribution .ipa/.aab, fixing signing/provisioning-profile or entitlement errors, bumping CFBundleVersion/versionCode, filling App Privacy / Data safety forms or PrivacyInfo.xcprivacy, wiring Fastlane lanes, or diagnosing a store rejection. Do NOT use for general (non-mobile) release steps like changelogs, tagging, or backend deploys - use prepare-release instead; do NOT use for store listing titles, descriptions, or keywords - use app-store-copy instead.
What it loads into context
Frontmatter
154 tokensAlways available to the agent
SKILL.md
1,320 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore6,178 bytes
Read SKILL.md
# App Store Release Prep
Turn a mobile release into a reproducible pipeline - correct signing, monotonic build numbers, complete privacy declarations - so submission to App Store Connect or Play Console succeeds on the first upload.
## Workflow
1. **Establish identity and signing first; it is the most common blocker.**
- iOS: a distribution build needs a distribution certificate plus a provisioning profile whose app id, capabilities, and entitlements all match the build. Store and sync these with Fastlane `match` in a shared encrypted repo so every machine and CI sign identically. Verify the profile matches the bundle id and enabled capabilities before building. Note that distribution certificates expire after one year - track the expiry so it never surprises a release week.
- Android: sign with an upload key and enroll in Play App Signing so Google holds the app signing key. A lost upload key is recoverable via the Play Console; protect the app signing key regardless.
2. **Set both version numbers correctly.**
- Marketing version (`CFBundleShortVersionString` / `versionName`) is human-facing and may repeat across uploads.
- Build number (`CFBundleVersion` / `versionCode`) must strictly increase on every upload - including re-uploads of a rejected build. Automate it with `increment_build_number` or derive it from the CI build count so you never collide with an existing upload. Android's `versionCode` has a hard ceiling of 2,100,000,000 - avoid date-encoded schemes that burn through the range.
3. **Build the signed artifact through a defined lane.**async-communication
When your agent uses it
Converts sync meetings and vague pings into structured asynchronous messages - TL;DR up front, an explicit ask with owner and deadline, and a default-action line that keeps work moving if nobody replies - plus team response norms per channel. Use when someone asks "how do I say this without a meeting", "rewrite this message so people actually respond", "help my remote team stop living on calls", or "draft an async decision doc". Do NOT use for the recurring exec status update itself - use eng-status-rollup instead - for extracting actions from a meeting that already happened - use meeting-notes-to-actions instead - or for designing the meeting you do keep - use meeting-agenda instead.
What it loads into context
Frontmatter
155 tokensAlways available to the agent
SKILL.md
1,554 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,114 bytes
Read SKILL.md
# Async Communication
Async lets distributed teams move without living on calendars, but the trade is real: it demands writing clear enough that the reader can act without a reply loop. The costly failure is the dead thread - a message with no explicit ask, no deadline, and no default, which stalls the work while everyone assumes someone else will respond. Every rule below exists to prevent that thread.
## Operating procedure
### Step 1: gather inputs
- What the sender actually wants from the reader - a decision, a review, an FYI, or an action. If they cannot name it in one sentence, that is the first fix.
- Who must act, who merely needs to know, and the real deadline.
- The team's time-zone spread and existing channel conventions, if any.
- The message or meeting being replaced, when rewriting.
Label assumptions as assumptions ("assuming Legal doesn't need to sign off on this").audience-targeting
When your agent uses it
Builds and refines audience and segment targeting for paid campaigns - first-party audience construction, lookalike expansion, exclusion lists, platform sizing floors, and saturation diagnosis - and delivers a filled targeting matrix per campaign. Use when someone asks "who should I target with these ads", "why is my ad set stuck in learning", "my CPMs keep rising but conversions are flat", "should I use a 1% or 5% lookalike", or is launching a campaign, diagnosing poor reach quality, or expanding into a new customer segment. Do NOT use for auditing an entire paid program's structure and spend efficiency - use paid-acquisition-audit instead; do NOT use for defining the ideal customer profile itself - use icp-persona-builder instead; do NOT use for market segmentation strategy outside paid media - use segmentation-strategy instead.
What it loads into context
Frontmatter
188 tokensAlways available to the agent
SKILL.md
1,739 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: targeting matrix
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore8,027 bytesSKILL.mdcore8,027 bytes
Read SKILL.md
# Audience Targeting
A perfect ad shown to the wrong audience converts poorly, and the most common budget leak in paid media is not bad creative - it is unexcluded customers, oversized lookalikes, and ad sets too small to ever exit the learning phase. This skill defines who sees the ad, in what order audiences get built, and when an audience is exhausted. Targeting work begins before creative work and gets revisited monthly as audiences saturate.
## Operating procedure
### Step 1: Gather inputs
Collect before recommending any audience. Label estimates as guesses.
- Platform(s) and monthly budget per platform.
- Campaign objective: acquisition, retargeting, or retention.
- First-party assets available: CRM customer list (with LTV field if it exists), site traffic volume by page type, email engagement data.
- Target CAC or CPA, and the conversion event being optimized.big-purchase-decision
When your agent uses it
Evaluates a major purchase with total cost of ownership, opportunity cost, affordability red lines, and a cooling-off rule, producing a scored buy/wait/walk verdict. Use when someone asks "should I buy this car", "can I afford this", "is this purchase a good idea", or is about to finance anything with a monthly payment. Do NOT use for fitting an approved purchase into the monthly plan - use budget-builder instead; for whether the purchase should outrank debt payoff or savings - use financial-planner; for home-purchase retirement trade-offs - use retirement-projection.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,817 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Purchase decision scorecard
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,672 bytes
Read SKILL.md
# Big Purchase Decision
Big purchases feel like one-time decisions but are almost always multi-year financial commitments. The costly mistake this skill prevents is deciding on sticker price and monthly payment - the two numbers sellers control - instead of total cost, opportunity cost, and affordability. Analyze before signing, because afterward the analysis is just regret.
## Operating procedure
### Step 1: Confirm the purchase clears the threshold
Apply the full framework to any single purchase above **1% of annual take-home income** (for most households, $500-$1,500). Two lighter rules cover everything below it:
- Below the 1% threshold but still discretionary and unplanned: apply the **24-hour rule** - sleep on it once. Most impulse purchases do not survive one night.
- A useful ceiling for guilt-free discretionary buys: anything under **1% of net worth** rarely warrants analysis for households with positive net worth; above that, it does.
Non-urgent purchases above the 1%-of-income threshold get the **30-day rule**: wait 30 days before buying. Most impulse-driven desires fade within that window; a desire still strong at day 30 that also passes the financial tests below is a considered want, not an impulse.bloodwork-explainer
When your agent uses it
Translates common lab panels - CBC, CMP, lipid panel, thyroid, hormones, HbA1c - into plain language, flags values against the lab's own reference ranges, and produces a prioritized question list to bring to a physician. Use when someone asks "what does my bloodwork mean", "is my ALT high", "explain my lipid panel", or "what should I ask my doctor about these results". General wellness education, not medical advice - it flags and explains, never diagnoses or prescribes. Do NOT use to summarize a medical chart or clinical document - use clinical-summary instead; for diet changes motivated by results, hand off to nutrition-planner.
What it loads into context
Frontmatter
156 tokensAlways available to the agent
SKILL.md
1,651 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,799 bytesSKILL.mdcore7,799 bytes
Read SKILL.md
# Bloodwork Explainer
Most people do one of two harmful things with lab results: self-diagnose from a single out-of-range value, or file the report away without understanding what it flagged. This skill does neither. It converts a lab report into plain-language explanations, honest flags, and a question list that makes the next doctor visit dramatically more productive. It never diagnoses, never recommends treatment, and never overrides a physician.
## Operating procedure
### Step 1: Gather inputs
Collect before interpreting anything. Missing context is the main source of bad interpretation.
1. The results themselves, including the reference range printed next to each value. Reference ranges are lab-specific - different labs use different assays and populations, so never substitute a generic range from memory for the one on the report. If the user gives values without ranges, ask for the ranges; label any range you supply yourself as a generic approximation, not the lab's.
2. Age and sex (many ranges differ by both).
3. Current medications and supplements (statins, biotin, creatine, and many others move markers).
4. Fasting status at the draw (glucose and triglycerides are meaningless to compare against fasting ranges otherwise).board-management
When your agent uses it
Runs a company board as an instrument - cadence, a no-surprises pre-wire, a board pack shipped days ahead, and meetings that end in decisions with owners. Use when someone asks "how do I prepare for my board meeting", "what goes in a board deck", "my board meetings feel like theater", "how often should we meet", or "how do I handle a difficult board member". Applies to venture-backed and private company boards; nonprofit and association boards follow different fiduciary and consensus norms, so adapt with care there. Do NOT use for writing the monthly investor update - use investor-update-writer instead - or for general internal meeting agendas - use meeting-agenda instead.
What it loads into context
Frontmatter
154 tokensAlways available to the agent
SKILL.md
1,713 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Advisory boards (distinct from governance)
- Board pack skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,478 bytes
Read SKILL.md
# Board Management
A board meeting is not a status update; it is the highest-leverage governance and strategy session a CEO gets, and the costly failure mode is predictable - a pack built the night before, a meeting spent narrating slides, and a director surprised by bad news in the room. A surprised board member becomes a defensive board member, and defensiveness compounds across quarters. This skill turns the board into a working instrument: pre-wired, decision-oriented, and useful between meetings.
## Operating procedure
Follow the steps in order. The pre-wire (Step 4) must come after the pack is drafted and before the meeting, because you cannot align directors on a decision they have not seen framed.
### Step 1: Gather inputs
Collect these before drafting anything. If a number is a guess, label it a guess.
1. Stage and cadence - current meeting frequency and stage of company. Default cadence: monthly for early stage, stretching to every 6-8 weeks as the company scales.
2. Board composition - each member, their affiliation (founder, investor, independent), and any known tensions or dominant voices.book-proposal
When your agent uses it
Writes non-fiction book proposals that sell to agents and acquiring editors - an overview hook answering why this book and why now, author platform with real numbers, target market, honest recent comps, a marketing plan, and a chapter outline that proves a full book exists. Use when an author asks "write my book proposal", "what comps should I use", "how do I pitch my book to an agent", or has a manuscript idea and needs the sales document publishers require. Do NOT use for writing the book's actual scenes or chapters - use ghostwriter instead; for pitching podcast appearances to promote the book, use podcast-pitch.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
767 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- Standard sections
- The hook
- Comps discipline
- Platform honesty
- Chapter outline quality
- Anti-patterns
- Output
Files
SKILL.mdcore3,501 bytes
Read SKILL.md
# Book Proposal
You write non-fiction book proposals - the sales document agents and acquiring editors read before they ever see a manuscript. A proposal sells the book's idea, market, and the author's ability to reach readers.
## Process
1. Gather: the book's core idea, the author's credentials and platform, the intended reader, comparable titles, and a rough table of contents.
2. Lead with the hook - why this book, why now, why this author.
3. Assemble the standard proposal sections.
## Standard sections
1. **Overview / hook** (2-4 pages) - the most important section. Open like the book itself: a compelling story, statistic, or provocation. Articulate the big idea, the problem it solves, and the promise to the reader. An editor decides here.
2. **About the author** - credentials, why YOU can write this. Authority and access matter.brand-guidelines
When your agent uses it
Writes a brand guidelines document - personality traits with is/is-not definitions, voice principles, a tone-by-context matrix, writing mechanics, a say-this-not-that word list, and visual usage rules - all made concrete with do/don't examples. Use when someone asks "write our brand guidelines", "document our voice and tone", "make a style guide the whole team can follow", or teams keep producing off-brand copy and design. Do NOT use for inventing the brand name itself - use brand-naming instead; for organizing messaging by audience and claim, use messaging-hierarchy.
What it loads into context
Frontmatter
135 tokensAlways available to the agent
SKILL.md
822 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core principle
- Process
- Sections
- Rules for the rules
- Anti-patterns
- Output
Files
SKILL.mdcore3,761 bytes
Read SKILL.md
# Brand Guidelines Doc
You write brand guidelines: the reference that keeps everyone - writers, designers, support, partners - sounding and looking like one coherent brand. Good guidelines are specific and usable, not abstract poetry.
## Core principle
A guideline is only useful if it changes what someone writes or designs today. Replace vague adjectives with concrete rules and examples. "Be friendly" is useless; "Use contractions and address the reader as 'you'; never use corporate jargon like 'leverage' or 'synergy'" is usable.
## Process
1. Gather: the brand's mission/values, audience, personality, and any existing materials to derive patterns from.
2. Define the verbal identity (voice, tone) and document visual principles.
3. Make everything concrete with do/don't examples.
brand-naming
When your agent uses it
Generates twenty-plus brand or product name candidates across five naming archetypes, scores them against weighted criteria without averaging, and runs linguistic and memorability safety checks. Use when someone asks "help me name my startup", "is this a good product name", "generate name ideas for this feature", or is auditing an existing name for distinctiveness, strategic fit, and risk. Not a substitute for professional trademark clearance. Do NOT use for crafting the positioning the name must express - use positioning-statement instead; for briefing the visual identity around a chosen name, use logo-brief-writer.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
1,534 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,965 bytes
Read SKILL.md
# Brand Naming
Naming is a strategic decision with long-term consequences: a name shapes perception, affects trademark viability, influences SEO, and is one of the hardest things to change once established. The costly failure is falling in love with the first clever candidate, then discovering after the logo, domain, and launch that it is unavailable, unpronounceable in a key market, or means something unfortunate in Portuguese.
## Operating procedure
The order is deliberate: criteria before generation prevents post-hoc rationalization, and volume before evaluation prevents premature convergence.
### Step 1: Gather inputs
- What is being named: company, product, or feature. The stakes and criteria weights differ.
- Category and the competitors' names - distinctiveness is only measurable against the actual field.
- Primary target language plus the top 3 secondary language markets.
- Constraints: domain budget, must it work as a verb, must it fit an existing naming system.budget-builder
When your agent uses it
Builds a zero-based monthly budget from take-home income using the 50/30/20 framework with explicit adaptation rules and sinking funds for irregular expenses. Use when someone says "help me make a budget", "I don't know where my money goes", "my spending feels out of control", or after an income change or move. Do NOT use for choosing what to fund first across debt, savings, and investing - use financial-planner instead; for sequencing multiple debts - use debt-payoff-planner; for sizing the cash cushion - use emergency-fund-planner; for company or campaign budget variance reporting - use budget-vs-actual instead.
What it loads into context
Frontmatter
146 tokensAlways available to the agent
SKILL.md
1,579 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,012 bytes
Read SKILL.md
# Budget Builder
A budget is not a restriction on spending - it is a spending plan made in advance. The failure this skill prevents is the budget that dies in month two: built on gross income, blind to irregular expenses, and never reviewed. Build one that survives contact with real life.
## Operating procedure
### Step 1: Gather inputs
Collect before allocating anything. Label guesses as guesses.
1. Monthly take-home income - net dollars that land in the bank, all regular sources (salary, side income, rental). If income varies, use the average of the last three months and adjust upward only when a higher month is confirmed.
2. Fixed essentials with amounts: rent or mortgage, utilities, insurance, minimum debt payments, transportation to work.
3. Variable essentials: groceries, fuel, medical. Pull from 2-3 months of actual statements, not memory - self-reported spending typically undershoots reality by 10-20%.
4. Wants: dining out, subscriptions, hobbies, entertainment. Statements again, not memory.budget-pacing
When your agent uses it
Paces and reallocates ad budget across channels through the month to hit both CAC and volume targets, with daily tolerance bands, a reallocation reserve, and escalation red lines. Use when someone asks "are we pacing to spend the budget", "we're overspending and it's only the 12th", "should I move budget from Meta to Google mid-month", "how do I spend the rest of the budget without wrecking CAC", or is managing live campaign spend at month start or month end. Do NOT use for explaining why last month's spend missed plan - use budget-vs-actual instead - or for auditing whether the channels themselves are worth funding - use paid-acquisition-audit instead.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,665 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: mid-month pacing check
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,662 bytes
Read SKILL.md
# Budget Pacing
Budget pacing is spending the right amount at the right time to meet both volume and efficiency targets. Overspending early depletes budget before the high-value conversion windows late in the month; underspending quietly wastes the month and the pipeline it was supposed to buy. The expensive failure this skill prevents is discovering on day 25 that the month is unrecoverable - every rule below exists to force the correction inside 48 hours instead.
## Operating procedure
Order matters: targets must exist before monitoring means anything, and monitoring must be daily before reallocation is safe.
### Step 1: Gather inputs
Collect before the month begins. Label estimates as guesses and tighten them with the first week's actuals.
1. Total monthly budget and the CAC target (or target range) per channel.
2. Volume target: conversions, leads, or purchases expected for the month.budget-vs-actual
When your agent uses it
Structures a budget-vs-actual variance analysis that isolates root causes - price/volume/mix decomposition, timing vs structural expense buckets, a materiality screen, and reforecast flags - instead of restating numbers. Use when someone asks "why did we miss budget", "write the variance commentary for the board deck", "explain this expense overrun", or is closing the month and owes narrative on the P&L. Do NOT use to build the budget or plan itself - use budget-builder instead; do NOT use for a full driver-based forecast model - use fpa-model instead; do NOT use for cash timing and runway questions - use cash-flow-forecast instead.
What it loads into context
Frontmatter
160 tokensAlways available to the agent
SKILL.md
1,658 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,205 bytes
Read SKILL.md
# Budget vs. Actual Variance Analysis
Variance analysis is diagnostic, not descriptive. A report that says revenue was 8% below budget is not analysis. Analysis says why, which team or product drove it, whether it is recoverable, and what it means for the full-year forecast. The costly failure is a wall of explanations for immaterial noise that buries the one structural miss the leadership team needed to act on.
## Operating procedure
### Step 1: Gather inputs
1. The actuals and budget at the same level of account granularity, for the period and YTD.
2. Total monthly revenue budget (it sets the materiality floor).
3. For revenue lines: budgeted and actual units and price by segment, if available. If not, note that decomposition will be directional and label it as such.
4. Known one-offs and timing shifts already identified by the accounting close (pair with month-end-close for the close itself).
5. The audience: internal ops review or board deck. Default: board-ready.
build-on-agent-native
When your agent uses it
Build an application on Builder.io's open-source Agent-Native framework (@agent-native/core), where the agent acts inside the app as a peer to the UI, sharing one set of actions and one database with the interface. Covers actions as the single source of truth, SQL-backed shared state, real-time sync, context-awareness, agent-to-agent calls, automations, the skills convention, and the CLI and template workflow. Use when someone says "build an agent-native app", "use the agent-native framework", "defineAction", "@agent-native/core", "clone an agent-native template", "make the agent a peer to my UI", or "an app where the agent and the UI call the same actions". Do NOT use for authoring generic SKILL.md files (use skill-creator), a standalone MCP server unrelated to this framework, or plain Next.js/React scaffolding with no agent runtime.
What it loads into context
Frontmatter
206 tokensAlways available to the agent
SKILL.md
1,691 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to reach for it
- Core primitives
- Start from a template, don't scaffold
- Read version-matched docs - do not trust memory
- Deliverable
- Quality bar
- Don't
Files
SKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytesSKILL.mdcore7,496 bytes
Read SKILL.md
# Build on Agent-Native
Agent-Native (open-source, MIT, by Builder.io - `@agent-native/core`) is a framework for apps where the agent acts **inside** the product, not in a chat box beside it. The agent and the UI are equal citizens of one system: the same action that a button calls, the agent calls as a tool, over the same database and shared state. Use this skill to build on it correctly instead of bolting a chatbot onto a normal app.
## When to reach for it
Choose Agent-Native when the agent needs to *do real work in the product* - read and mutate app data, edit the same document a human is editing, coordinate with other agents - and you want one implementation serving UI, agent, HTTP, MCP, A2A, and CLI. If you only need a chat assistant next to an otherwise-normal app, you do not need this framework; say so.
## Core primitives
**Actions are the single source of truth.** Define each operation once with `defineAction` (a Zod schema plus a `run` function). That one definition is callable from the agent as a tool, from the UI via `useActionQuery` / `useActionMutation`, and from HTTP, MCP, A2A, and the CLI (`pnpm action <name>`). Do not add `/api/*` pass-through routes whose only job is to call an action - call the action directly.
**Keep the action surface small and orthogonal.** Every agent-exposed action is a tool in the model's context window, and more tools degrade tool-selection quality - past roughly 15-20 agent-visible tools, expect noticeably worse selection, so consolidate before you get there. Prefer one patch-style `update-<thing>` over many per-field actions (an app exposing 40 per-field setters where 5 patch actions would do is the canonical smell). Hide UI-only or programmatic actions from the model with `agentTool: false` (still callable from the UI/HTTP, just not a tool the agent sees). Note that `agentTool: false` (hidden from the model entirely) is *not* `toolCallable: false` (blocks only the sandboxed extension iframe bridge) - use the latter for high-blast-radius operations, not for trimming the tool list.
business-rule-extractor
When your agent uses it
Produces a documented inventory of the implicit business rules, edge cases, and bug-as-feature behaviors that tangled code encodes, with real input-to-output examples, so a rewrite preserves behavior. Use when you are about to rewrite, port, or replace legacy code whose only specification is the source, and you see policy buried in conditionals, magic numbers, hardcoded dates, or per-customer special cases. Do NOT use when you need an executable safety net before refactoring - use characterization-test-writer instead; do NOT use when you need to find service boundaries in a monolith - use monolith-decomposer instead.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,016 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,963 bytesSKILL.mdcore4,963 bytesSKILL.mdcore4,963 bytes
Read SKILL.md
# Business Rule Extractor
Turn undocumented legacy logic into a written rule inventory so a rewrite preserves behavior instead of silently changing it. The source code is the spec, and most of it is accidental - your job is to surface what it actually does, not what it should do.
## Workflow
1. **Scope the extraction target.** Identify the exact module, files, and entry points to be rewritten. List every external system, shared table, global, and session state the logic touches - a rule split across files is the one most likely to break on rewrite.
2. **Mine rules from conditionals and constants.** For each branch, guard, magic number, string literal, and validation, ask "what business reality does this encode?" `amount > 10000` is a large-transaction threshold; a hardcoded date is a regulation start; a branch on one customer ID is a negotiated exception. Magic numbers are almost always undocumented policy. Treat any constant that appears in two or more places as a single rule with duplication risk - note every occurrence, because a rewrite that updates one copy and misses the other creates a divergence bug.
3. **Record order and interaction.** Capture precedence (which discount wins when two apply), short-circuits, fall-through defaults, and exact boundary handling of null, empty, negative, and `>` vs `>=`. Evaluation order is itself a rule; note every place an input path forks.
4. **Separate intended rules from bug-as-features.** Code alone cannot tell you which behaviors are deliberate. Flag each questionable behavior and confirm against domain experts, support tickets, or production data - aim for at least two independent sources (e.g. one expert plus production evidence) before labeling a behavior intended. Document bug-as-features explicitly so the rewrite team decides consciously, not in an incident.
5. **Capture each rule as input-to-output examples.** For every rule, record concrete cases from real data, especially at boundaries - a minimum of 3 cases per rule (below, at, and above each boundary), and 10+ for rules with interacting conditions. A table of cases removes the ambiguity that "applies a tiered discount" leaves behind.
6. **Hand off open questions.** Before declaring a path dead, check production logs or traffic over a window that covers the business cycle - 90 days is a practical floor, a full year for anything with annual/seasonal triggers. Mark provably dead paths as dead rather than reverse-engineering intent. List behaviors you could not classify as explicit questions for domain owners.
## Deliverable
buyer-nurture-sequence
When your agent uses it
Keeps not-yet-ready homebuyers warm for months with timeline-based segmentation, listing alerts with a why-this-one note, a monthly market-update touch, the pre-approval nudge, and re-engagement triggers. Use when an agent says "keep my buyer leads warm", "these buyers aren't ready yet", "build my buyer drip", or "leads keep going cold and buying with someone else". Do NOT use for generic marketing email drips - use email-drip-builder instead - or for post-open-house first contact, which is open-house-follow-up.
What it loads into context
Frontmatter
124 tokensAlways available to the agent
SKILL.md
2,206 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Segmentation and cadence
- The listing-alert-plus-context play
- The monthly market-update touch
- The pre-approval nudge
- Re-engagement triggers
- Thresholds
- Deliverable
- Do NOT
- Quality bar
- Escalation and compliance
Files
SKILL.mdcore9,907 bytesSKILL.mdcore9,907 bytes
Read SKILL.md
# Buyer Nurture Sequence
Most buyer leads are not ready today - they are ready in four to ten months, and they hire whichever agent is still usefully present when readiness arrives. The costly mistake this skill prevents is the binary treatment of leads: work the hot ones, forget the rest. The forgotten ones buy anyway, with someone else, and the agent re-pays acquisition cost for a lead they already owned.
Worked example throughout: Priya, a residential agent in a suburban market, 14 transactions last year at a $485,000 average sale price, building toward 24. Her open houses (see open-house-follow-up, the upstream skill that feeds this one) produce 3-5 nurture entries each; over a year that is a pool of roughly 60-80 not-yet-ready buyers. Converting even 8-10 of them is most of the gap between 14 and 24 transactions - at her $485k average price, each converted nurture lead is a five-figure commission event that cost nothing new to acquire.
## Operating procedure
1. **Segment every incoming lead by stated timeline** before anything else - cadence, content, and the call schedule all depend on it, so an unsegmented lead cannot be nurtured, only spammed. Intake carries the handoff notes from open-house-follow-up (or wherever the lead originated): timeline, financing status, criteria, and anything they said.
2. **Set up the listing-alert-plus-context play** for every segment. This is the core mechanic.
3. **Schedule the monthly market-update touch** for the whole pool.
4. **Place the pre-approval nudge** at the segment-appropriate moment.
5. **Wire the re-engagement triggers** so behavior, not the calendar, promotes leads between segments.
6. **Review the pool monthly:** re-segment anyone whose behavior or replies changed, and prune only hard opt-outs - never prune for silence.buying-signal-tracker
When your agent uses it
Use when you need to decide WHO IN YOUR ICP TO HIT NOW based on timing and trigger events rather than static fit - "who should I reach out to this week", "what's a good reason to reach out", "track buying signals", "find trigger events", "champion changed jobs", "they just raised a round", "new VP just started", "they're hiring for X", "they installed a competitor", "intent data spiked", "why now opener", "warm up our outbound", "score and prioritize accounts by signal". Workflow: enumerate the six signal categories (company triggers, people/job-change triggers, hiring reqs, technographics, third-party intent, first-party engagement), map each signal to a "why now" opening line and the play/sequence it should fire, then rank accounts with a fit×signal scoring matrix so you work the hottest first. Do NOT use for defining who fits in the abstract (firmographics, personas) - use [[icp-persona-builder]]; do NOT use for the tool mechanics of pulling/saving signals - use [[apollo-prospecting]].
What it loads into context
Frontmatter
238 tokensAlways available to the agent
SKILL.md
2,739 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The workflow
- Signal → play mapping template
- Worked example: champion job change
- Scoring artifact
- Deliverable
- Quality bar
- Common failure modes
Files
SKILL.mdcore11,091 bytesSKILL.mdcore11,091 bytesSKILL.mdcore11,091 bytes
Read SKILL.md
# Track Buying Signals & Trigger Events
Fit tells you WHO could buy; signals tell you WHO TO CALL TODAY. The single biggest lever in outbound is not a better list or a cleverer line - it is timing. A perfectly-fit account contacted in a dead quarter ignores you; the same account contacted the week a new VP inherits a broken process replies. ICP is static and slow-moving ([[icp-persona-builder]]); signals are dynamic and perishable. This skill is about catching the perishable window.
The common trap is treating all signals as equal and "spraying" the whole ICP whenever any signal fires. Signals vary wildly in strength: a champion who moves to a new in-ICP company is near-deterministic (someone who already bought you, now with budget and a mandate). A third-party intent surge is directional noise at best. Weight them, decay them by recency, multiply by fit - then work the top of the ranked list, not the whole thing.
## When to use this skill
- You have an ICP and a list but no sense of which accounts are *ripe* right now.
- A specific event just happened (funding, exec hire, layoff, job change) and you need the right "why now" angle and play.
- Your outbound is high-volume, low-reply and you suspect it's a timing problem, not a copy problem.
- You want to build a repeatable account-prioritization score instead of working alphabetically.
Do NOT use this to define firmographics or personas - that's [[icp-persona-builder]]. Do NOT use it for the click-path of saving signals in a tool - that's [[apollo-prospecting]].
caching-strategy
When your agent uses it
Designs multi-layer caching - CDN, application, and database - choosing the pattern, TTL, and invalidation plan per data type, and produces a cache-decision table plus stampede protection. Use when someone asks "should I cache this", "Redis or CDN for this endpoint", "how do I invalidate this cache", "why is my cache hit rate so low", or is bolting Redis onto a slow service. Do NOT use for sizing or tuning database connection pools - use connection-pool-tuner instead - and do NOT use for frontend bundle, image, or Core Web Vitals work - use web-performance instead.
What it loads into context
Frontmatter
142 tokensAlways available to the agent
SKILL.md
1,986 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: cache-decision table
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,198 bytes
Read SKILL.md
# Caching Strategy
A cache is a correctness liability you accept in exchange for latency, so every cache needs an explicit answer to "how stale can this be?" and "how does it get invalidated?" before it ships. The costly mistake this skill prevents is the write-only cache: a layer that adds a network hop and an invalidation bug surface while its hit rate quietly sits below 50 percent, making the system slower and wronger at the same time.
## Operating procedure
Work per data type, not per system - "cache the API" is not a decision, "cache the product listing for 60s at the CDN" is.
### Step 1: Gather inputs
Collect these for each candidate data type. If a number is an estimate, label it a guess and refine from production metrics later.
1. Read:write ratio. How many reads per write to the same key.
2. Recompute cost. Latency of a cache miss (DB query, computation, upstream call).candidate-outreach-personalizer
When your agent uses it
Drafts short, honest recruiting outreach and InMail messages to a passive candidate, anchored on one researched, candidate-published professional fact, for a human recruiter to review and send. Use when someone asks "write an InMail to this engineer", "personalize this sourcing message", "draft outreach from this profile", or is contacting a passive candidate about a specific role. Do NOT use for sales or founder cold email to prospects - use cold-email-craft. Do NOT use for designing the multi-touch cadence, channel mix, and follow-up timing around the message - use outreach-sequence-designer. Do NOT use for writing the job description itself - use job-description-writer.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,526 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast pair
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,164 bytesSKILL.mdcore7,164 bytes
Read SKILL.md
# Candidate Outreach Personalizer
Draft a short, honest, easy-to-decline recruiting message that earns a reply because it is unmistakably about this candidate and this role. The costly mistake this prevents is the template blast: passive candidates receive a steady stream of "your background is impressive" messages, and one more of those doesn't just get ignored - it burns the company's name with exactly the people it wants to hire. One researched fact beats three generic compliments every time.
## Operating procedure
### Step 1: gather inputs
- The candidate's public professional profile (URL or pasted summary).
- The role: title, level, scope, team, location/remote policy, and the pay range where disclosure is required.
- The sender: who the message is from (recruiter, hiring manager, founder) and their real connection to the role.
- Any true, specific reason this candidate surfaced (a project, a talk, a repo - not "the algorithm").
If the role's level or pay range is unknown, ask; guessing either produces a message that misleads.cap-table-manager
When your agent uses it
Use when a founder needs to understand or model ownership and dilution. Triggers on "model my cap table", "how much will I be diluted", "option pool", "SAFE conversion", "post-money vs pre-money", "what do I own after this round", "pro rata". Turns equity from a black box into a model the founder controls.
What it loads into context
Frontmatter
79 tokensAlways available to the agent
SKILL.md
693 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The mechanics that actually move ownership
- Build the model
- Guardrails
- Anti-patterns
- Deliverable
Files
SKILL.mdcore2,996 bytes
Read SKILL.md
# Cap Table Manager
Founders lose more value to cap-table mistakes they did not model than to
valuation they negotiated badly. This skill makes dilution legible: you should
be able to state what you own after the next two rounds, and why.
## When to use this skill
Use it during [[fundraising-stage-selector]] to pressure-test dilution, before
signing any SAFE or term sheet, and whenever someone says "we'll just expand the
pool." Pair instrument mechanics with [[safe-vs-priced-round]].
## The mechanics that actually move ownership
captions-from-transcript
When your agent uses it
Produce an accurate, properly timed caption track (SRT or WebVTT) from a video's audio - transcribing or aligning to the voiceover script, timing cues to speech, and enforcing line-length and reading-speed rules so captions are readable and in sync. Use when someone says "generate captions", "make subtitles from the audio", "transcribe and caption this", "export an SRT or VTT", "the captions are out of sync", or "the subtitles flash by too fast to read". Do NOT use to burn-in, style, reframe, or position an existing caption track for a platform (9:16, brand styling, safe areas) - that is social-video-formatter; do NOT use to animate text word-by-word as a motion graphic - that is kinetic-typography; do NOT use to write the spoken script in the first place - that is narration-script.
What it loads into context
Frontmatter
201 tokensAlways available to the agent
SKILL.md
946 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Source the most accurate text first
- Time the cues to speech
- Line and reading rules
- Choose the format
- Clean vs verbatim
- QA the sync
- Hand off
- Don't
Files
SKILL.mdcore4,065 bytes
Read SKILL.md
# Captions From Transcript
Captions are not optional polish - they are accessibility, they hold viewers watching muted, and for tutorials they reinforce exact UI terms. This skill produces a **clean, accurately timed caption file**. Styling and burn-in happen downstream.
## Source the most accurate text first
Accuracy comes from where you start:
- **If a narration script exists, use it as ground truth.** It is already correct on technical terms and UI labels. Align it to the audio rather than re-transcribing from scratch.
- **Otherwise transcribe the audio** (Whisper, a platform's auto-caption, or any ASR), then **correct it against the video**. ASR reliably mangles product names, code, and acronyms - fix every one to match what is on screen exactly.
Never ship raw ASR output. The errors are always in the highest-value words.
## Time the cues to speechcarousel-scripter
When your agent uses it
Scripts a multi-slide Instagram or LinkedIn carousel as a hook slide, one-idea-per-slide value frames, and a single-CTA closing slide, with per-slide copy and design direction. Use when the user asks to write, script, or outline a carousel, turn a post/idea/listicle into swipeable slides, or build an Instagram carousel or LinkedIn document (PDF) post.
What it loads into context
Frontmatter
99 tokensAlways available to the agent
SKILL.md
802 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Platform calibration
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore3,497 bytes
Read SKILL.md
# Carousel Scripter
Turn one idea into an ordered slide deck where slide one earns the swipe and every slide after earns the next.
## Workflow
1. Confirm the format fits. A carousel earns its keep on multi-step, list, or framework content. If the idea is a single quick tip or a time-sensitive moment, say so and recommend a Reel, single image, or text post instead - do not script a carousel for a one-liner.
2. Confirm platform, audience, and the one outcome the post should drive (follow, save, comment, click). Default to Instagram if unstated and ask once.
3. Draft 2-3 cover (slide 1) options. Each is a short headline that stops the scroll and promises a payoff - a number ("7 mistakes…"), a contrarian claim, or a clear transformation - plus a swipe cue. Spend disproportionate effort here.
4. Lay out the spine: Hook (slide 1), Stakes / why-it-matters (slide 2), Value slides (3 through n-1, one point each), CTA (final slide). Target 6-10 slides total.
5. Write each value slide as one point: a punchy headline (the layer skimmers read) plus 1-2 short support sentences. Front-load the point in the first words. Number each slide.
6. End most value slides with a tiny open loop or forward cue so the reader wants the next slide.
7. Write the closing slide with exactly one ask, tied to the value delivered ("Save this before your next launch").
8. Add a one-line design direction per slide (what the visual shows), since copy and layout are inseparable.
9. Write a caption that restates the hook and adds context the slides could not hold.
case-study-builder
When your agent uses it
Turn a customer win into a credible case study - interview-driven raw material, a challenge-solution-results arc, and a quantified results section with real customer quotes. Use when someone asks "write a case study from this customer call", "turn this win into a customer story", "what should I ask the customer in the interview", or "make our results section stronger". Do NOT use for collecting short standalone quotes or review snippets - testimonial capture is upstream raw material for this skill. Do NOT use for packaging finished case studies into battlecards and sales collateral - use sales-enablement-kit instead.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
1,629 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Case study skeleton template
- Results section: bad vs good
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,374 bytes
Read SKILL.md
# Case Study Builder
A case study sells by proof, not adjectives - the customer and the numbers do the talking. The costly mistake is writing it backwards: drafting marketing copy first and asking the customer to bless it, which produces approved-by-legal mush with no quotes worth reading and no numbers worth citing. Get the raw material first; the story is assembled, not invented.
## Operating procedure
The order matters: interview before outline, numbers before draft. A draft written before the interview anchors on your product's talking points instead of the customer's actual story.
### Step 1: Gather inputs
1. The customer: company, industry, size, and the buyer persona this case study should persuade (default: mirror the customer's own profile - readers trust stories about companies like theirs).
2. Access: can you interview the customer live (best), send written questions (acceptable), or only mine existing calls/emails/reviews (weakest - flag the limitation)?
3. Any numbers already known: baseline metric, result metric, timeframe. Label estimates as estimates.
4. Distribution target: web page, PDF leave-behind, or sales deck slide (default: one-page web layout).cash-flow-forecast
When your agent uses it
Builds a rolling 13-week and 12-month cash flow forecast with runway view. Use when managing liquidity, planning for fundraising, preparing board materials, or stress-testing the business under downside scenarios.
What it loads into context
Frontmatter
58 tokensAlways available to the agent
SKILL.md
1,157 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Two Horizons, Two Approaches
- 13-Week Forecast Structure
- 12-Month Forecast Structure
- Runway Calculation
- Working Capital Levers
- Worked example: forecasting collections
- Deliverable
- Quality bar
- Common Mistakes
Files
SKILL.mdcore5,328 bytes
Read SKILL.md
# Cash Flow Forecast
A cash forecast is not a P&L on a different template. It tracks when cash actually moves, not when revenue is recognized or expenses are accrued. Precision on timing is the entire point.
## Two Horizons, Two Approaches
Run two forecasts in parallel. The 13-week forecast is a bottom-up, week-by-week view of actual cash in and out. It is operational. The 12-month forecast is a top-down driver-based model used for planning and runway. They should agree in the overlapping weeks - if they diverge, the 13-week is right and the 12-month needs recalibration.
## 13-Week Forecast Structure
For each week, project: collections from customers (use AR aging plus expected new bookings converted with your average days-to-collect), payroll runs (use the exact pay schedule), vendor payments (use AP aging plus expected new invoices), recurring SaaS and fixed costs (pull from your vendor contracts), and any scheduled debt or tax payments. Sum to net cash change per week, accumulate to ending cash balance. Flag any week where ending cash falls below your minimum operating reserve (typically one to two months of fixed costs).
## 12-Month Forecast Structure
category-page-copywriter
When your agent uses it
Write the intro and lower-page copy for an ecommerce collection or category page so it ranks for a commercial query while pushing shoppers toward a product fast. Use when someone asks "write copy for my ecommerce category pages", "write the intro blurb for my collection page", or "write SEO copy for my collection page below the product grid" - the above-grid intro, the H1, and the supporting content under the grid. Do NOT use for standalone marketing or campaign landing pages - use landing-page-copy instead; do NOT use for a single product's description - use product-description-writer instead.
What it loads into context
Frontmatter
139 tokensAlways available to the agent
SKILL.md
771 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
Files
SKILL.mdcore3,596 bytes
Read SKILL.md
# Category Page Copywriter
A category page has a harder job than a product page: it must rank for a broad commercial query while helping a shopper narrow toward a product fast. It cannot push products below the fold or read like keyword salad.
## Workflow
1. **Gather inputs.** Get the category name, the primary keyword and 1-3 secondary keywords, the products it contains (count and the axes they vary on), the buyer's main decision criteria, and the brand voice. If any are missing, ask before writing.
2. **Write a tight above-grid intro.** 2-3 sentences, ~40-60 words, placed above the product grid. State what the collection is and who it's for, and use the head keyword once, naturally. Never wall the products off behind a long essay.
3. **Set the H1.** One clear H1 containing the head keyword, phrased for a shopper, not stuffed.
4. **Write a "How to choose" block.** 3-4 decision criteria that route the shopper to the right subset ("Pick the 150 weight for high-output activity, the 250 for cold-weather layering"). Bold each criterion. This is the highest-value part of the page.
5. **Write the lower-page SEO content.** Below the grid, in short paragraphs under subheads, answer buying questions: how to choose, what differs between options, what fits which use case. This captures "best", "vs", and "for [use case]" long-tail intent through semantic coverage, not keyword repetition.
6. **Add internal links and FAQ copy.** In the lower content, link to relevant sibling categories and key products using descriptive anchor text. Draft 3-5 FAQ questions with concise answers (real shopper questions) for the page owner to publish.
7. **Stress-test the category.** If the category is thin or overlaps another, flag it before delivering copy (see Do NOT).
causal-inference
When your agent uses it
Selects and executes a credible causal identification strategy - RCT, natural experiment, difference-in-differences, regression discontinuity, instrumental variables, or matching - ranked by assumption strength, with a confounder checklist, falsification tests, and a defensible effect estimate. Use when someone asks "did X actually cause Y", "how do I measure impact without an A/B test", "is this correlation causal", or needs to defend an effect estimate from observational data. Do NOT use for designing the data-collection study itself (sampling, instruments, power) - use primary-research instead; for reading out an already-run A/B test, use ab-test-analyzer instead.
What it loads into context
Frontmatter
156 tokensAlways available to the agent
SKILL.md
1,978 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: identification-strategy choice
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore9,319 bytesSKILL.mdcore9,319 bytesSKILL.mdcore9,319 bytesSKILL.mdcore9,319 bytes
Read SKILL.md
# Causal Inference
A confident effect estimate built on a broken identification strategy is worse than no estimate: it triggers real decisions with fake evidence. This skill forces the choice of design before any regression is run, states the assumptions each design buys, and requires falsification tests before an effect is reported as causal.
## Inputs to collect
Gather these before choosing a design. If the user cannot supply one, propose a default and label it a guess.
1. **Treatment**: the intervention, precisely defined (who received it, when, at what intensity).
2. **Outcome**: the measured variable and its timing relative to treatment.
3. **Unit of analysis**: person, account, store, region - and how many units exist on each side.
4. **Assignment mechanism**: how units ended up treated - randomized, threshold rule, policy rollout, self-selection. This single fact determines which designs are available.
5. **Data availability**: pre-treatment periods (how many), untreated comparison units, candidate instruments, covariates.
6. **Decision at stake**: what action the estimate will drive, and how costly a wrong sign or 2x-off magnitude would be.changelog-generator
When your agent uses it
Generate or update a repository CHANGELOG.md in Keep a Changelog format from git history - grouping commits since the last tag into Added/Changed/Deprecated/Removed/Fixed/Security, rewriting them as user-facing entries, and recommending the semantic version bump. Use when someone asks "generate a changelog from the git log", "update CHANGELOG.md for this release", "what version bump does this release need", or "turn these commits into release notes". Do NOT use for writing benefit-first product announcement changelogs for end users from release notes - use changelog-writer instead; this skill produces the versioned CHANGELOG.md file that lives in the repo.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,463 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Format
- Worked contrast - translating a commit
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,666 bytes
Read SKILL.md
# Changelog Generator
A changelog exists so a human deciding whether to upgrade can answer "what breaks, what's fixed, what's new" in thirty seconds. The costly failure is dumping `git log` into a file: raw commit messages describe the code's history, not the release's meaning, and they bury the one breaking change under forty chores. This skill turns commit history into a Keep a Changelog-formatted CHANGELOG.md and a defensible version bump.
## Inputs to collect
Gather before writing: the repo (or the raw log), the last released tag (`git describe --tags --abbrev=0` - if there are no tags, the whole history is the first release), the existing CHANGELOG.md if any (append to its conventions rather than reformatting history), whether commits follow Conventional Commits (`feat:`, `fix:` prefixes make categorization mechanical; otherwise categorize by reading each diff's intent), and the project's versioning scheme (default: SemVer). Label any inferred categorization from an ambiguous commit as a guess for the maintainer to confirm.
## Operating procedure
Order matters: categorize before writing, and decide the version last, because the bump is a function of the finished category lists.
1. **Collect the range.** `git log <last-tag>..HEAD --oneline --no-merges`. Pull PR numbers from merge commits or squash-commit suffixes like `(#123)`.
2. **Drop invisible commits.** Remove chores, CI tweaks, test-only changes, dependency bumps, refactors, and formatting - unless one produced a user-visible result, in which case describe the result, not the work ("Reduced install size by 40%", not "Switched bundler").changelog-writer
When your agent uses it
Translates raw engineering release notes - commits, tickets, PR titles - into a user-facing changelog grouped as New, Improved, and Fixed, with a highlights section on top, benefit-first phrasing in the user's vocabulary, bugs described by their symptoms, and internal-only churn cut and flagged. Use when someone says "write the changelog for this release", "turn these commit messages into release notes users can read", "announce what shipped this week", or their changelog currently reads as internal jargon. Do NOT use for auto-generating a technical changelog from git history for developers - use changelog-generator instead.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
801 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core principle
- Process
- Categories (group entries)
- Writing each entry
- What to include vs. cut
- Structure of a release entry
- Tone
- Anti-patterns
- Output
Files
SKILL.mdcore3,630 bytes
Read SKILL.md
# User-Facing Changelog
You translate engineering release notes into a changelog real users want to read. Developers describe what changed; users care what it means for them.
## Core principle
Lead with the benefit, not the implementation. "We migrated to a new caching layer" means nothing to a user. "Pages now load about twice as fast" does.
## Process
1. Take the raw notes (commits, tickets, PR titles).
2. Group them by user impact, drop the invisible internal churn.
3. Rewrite each meaningful change in user language, benefit first.
channel-strategy
When your agent uses it
Picks, sequences, and scales the distribution channels that fit a company's economics, producing a scored channel matrix and a 12-month sequencing plan. Use when someone asks "which marketing channel should we focus on", "should we do paid or content or outbound", "our growth is stalling across channels", "when do we add a second channel", or "is this channel worth scaling". Do NOT use for auditing an existing paid-ads account - use paid-acquisition-audit instead - for the full go-to-market plan around the channels - use go-to-market-planner instead - or for modeling how channels compound into a growth model - use growth-model instead.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,614 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: channel matrix
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,136 bytes
Read SKILL.md
# Channel Strategy
Most startups die from lack of distribution, not lack of product, and the most common self-inflicted wound is spreading effort across ten channels before any single one works. At any given stage, one channel drives the majority of growth; the job is to find it, saturate it, and only then add the next. This skill produces the scored matrix and the sequencing plan that enforce that focus.
## Operating procedure
Work the steps in order - scoring before sequencing, sequencing before spend - because a sequencing plan built on unscored channels is a wish list.
### Step 1: Gather inputs
Collect these before scoring anything. Label guesses as guesses.
1. ACV (annual contract value) or average order value - this alone eliminates channels.
2. Buyer behavior - do buyers search, scroll, get referred, or get sold to?characterization-test-writer
When your agent uses it
Writes pinning and characterization tests that lock in the current behavior of untested legacy code - bugs included - before a refactor, so any later behavior change trips an alarm. Use when someone says "I need a safety net before refactoring this", "this module has no tests and I have to change it", "capture what this function does today", or the correct behavior is simply whatever the code does now. Do NOT use for writing tests for new behavior built red-green - use tdd-expert instead; do NOT use to decide which code is worth testing or to rank coverage gaps - use coverage-gap-finder instead; do NOT use to verify existing assertions are strong - use mutation-test-runner instead.
What it loads into context
Frontmatter
163 tokensAlways available to the agent
SKILL.md
1,426 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,581 bytesSKILL.mdcore6,581 bytesSKILL.mdcore6,581 bytesSKILL.mdcore6,581 bytes
Read SKILL.md
# Characterization Test Writer
Pin what legacy code currently does - bugs included - so a refactor that changes behavior fails loudly instead of silently. The costly mistake this prevents is the "clean" refactor that quietly changes an output nobody noticed was load-bearing, discovered weeks later in production. A characterization test records reality; it never judges it.
## Operating procedure
Order matters: the seam must exist before you can capture output, output must be captured before you can stabilize it, and coverage is only measurable once tests run.
### Step 1: Gather inputs
- The exact function/module about to change, and the refactor's blast radius - which code paths will the change touch? Only those need pinning.
- The dependency list: database, clock, RNG, network, filesystem, globals, environment. Each is a nondeterminism source to control in Step 4.
- Any recorded production inputs or logs available for replay. Label assumptions about "typical inputs" as guesses.
### Step 2: Find the smallest seamchurn-reduction
When your agent uses it
Diagnoses SaaS churn root causes through cohort analysis and a seven-category taxonomy, then builds segmented intervention playbooks ranked by frequency, revenue at stake, and addressability - including save-offer economics. Use when someone asks "why is our churn so high", "build a churn reduction plan", "should we offer discounts to cancelling customers", or when NRR/GRR is slipping and the team needs a diagnosis before prescriptions. Do NOT use for gym or fitness-studio member retention - use retention-and-churn-killer instead. Do NOT use for re-acquiring customers who already left - use win-back-campaign instead. Do NOT use for growing existing accounts - use expansion-revenue instead.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,815 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Metrics to track ongoing
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,380 bytes
Read SKILL.md
# Churn Reduction
Churn is the silent killer of SaaS - 5% monthly churn caps the business no matter how fast acquisition runs, because it compounds into LTV, NRR, and valuation. The costly mistake this skill prevents is prescribing before diagnosing: launching a save-offer program or a CS hiring spree against a blended churn number, when the real leak is bad-fit acquisition or failed onboarding that no discount can fix.
## Operating procedure
Diagnosis strictly precedes intervention. Steps 1-3 establish what is actually leaking; only then do playbooks get built.
### Step 1: Gather inputs
1. Monthly logo churn and revenue churn for the trailing 12 months. Track both - a small customer leaving and a whale leaving are different problems.
2. GRR and NRR. Gross revenue retention (churn + contraction only) isolates the leak; net revenue retention (includes expansion) can hide it. An NRR of 105% with GRR of 82% is a leaky bucket propped up by a few expanding whales.
3. Customer list with segment, plan, signup month, acquisition channel, ACV.
4. Cancellation reasons, exit interviews, and support history where they exist. Default: sparse - plan interviews in Step 3.circuit-breaker-builder
When your agent uses it
Wraps flaky upstream dependencies in circuit breakers, aggressive timeouts, and per-dependency bulkheads so a slow or failing service degrades gracefully instead of cascading into a full outage. Use when a slow or unavailable upstream is stalling your threads, outbound calls hang with no timeout, one dependency's outage is taking down unrelated traffic, or you are integrating a network call that can realistically be slow or down. Do NOT use when the goal is staying under a provider's request quota or handling 429s - use rate-limit-handler instead; do NOT use to size a database connection pool - use connection-pool-tuner instead.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,084 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,198 bytes
Read SKILL.md
# Circuit Breaker Builder
Protect a service from slow or unavailable upstreams: fail fast, fall back, and isolate the blast radius so one sick dependency can't drain your whole capacity.
## Workflow
1. **Set an explicit timeout on every outbound call.** Configure connect and read timeouts separately; never rely on library defaults (often minutes or infinite). A sane starting point: connect timeout of 1-2 seconds, read timeout anchored to the dependency's observed p99 latency plus modest headroom - not a round number pulled from the air. The total budget must be shorter than your own caller's timeout so you fail before they give up. A call with no timeout is a guaranteed resource leak under load.
2. **Wrap the call in a three-state breaker.** Closed: calls pass through. Open: calls fail instantly without touching the dependency. Half-open: a few probe calls test recovery. Implement these states explicitly per dependency, not a single global breaker.
3. **Trip on a rolling error rate, not a raw count.** Open the breaker when the error rate crosses a threshold over a sliding window with a minimum request volume. Widely-used defaults from Hystrix and resilience4j are a good anchor: trip at ≥50% errors over the window, but only once at least 10-20 requests have been observed in it. A raw failure count flaps at low traffic and trips on noise.
4. **Define a fallback per call site.** While open, return immediately: stale cache for reads, queue-for-later or accept-and-reconcile for writes, a sane default or degraded response where one exists, and a clean fast error only when there is no safe degradation. The point is to stop spending threads, connections, and latency on a call you expect to fail.
5. **Probe and recover from half-open.** After a cooldown - 30-60 seconds is a reasonable starting range - allow a limited number of probe calls (single digits, e.g. 3-5). If they succeed, close the breaker. If any fail, re-open and reset (and back off) the cooldown. Never flood the recovering dependency with full traffic the instant the cooldown expires.citation-tracker
When your agent uses it
Verifies that every citation is real, unretracted, and complete, rates source quality by evidence tier, and formats reference lists precisely in APA, MLA, Chicago, or IEEE. Use when someone asks "check my references", "format this bibliography in APA", "are these citations real", or "convert my reference list to IEEE". Do NOT use for checking whether the claims in the text are true - use fact-checker or claims-verifier instead. Do NOT use for writing the review itself - use literature-review or systematic-review instead.
What it loads into context
Frontmatter
125 tokensAlways available to the agent
SKILL.md
1,397 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytesSKILL.mdcore6,440 bytes
Read SKILL.md
# Citation Tracker
Ensure every citation is accurate, complete, and correctly formatted for the requested style. Accuracy comes before formatting - a beautifully formatted wrong citation is still wrong, and the costly failure this skill prevents is the fabricated or unverifiable reference that survives to publication and detonates the author's credibility on contact with one curious reader.
## Operating procedure
Verification precedes formatting, because formatting effort spent on a citation that turns out to be retracted or nonexistent is wasted twice.
### Step 1: Gather inputs
Collect: the reference list and the manuscript (both - cross-matching needs both sides), the target style, and the venue's house rules if any. If the style is unstated, ask: APA 7, MLA 9, Chicago (notes-bibliography or author-date), or IEEE. Never mix styles in one document, and for Chicago, pin down which variant before touching anything.
### Step 2: Inventory the required elements
claims-verifier
When your agent uses it
Decomposes a claim into checkable propositions, traces each through the citation chain to its primary source, appraises the evidence for method and independence, and issues a verdict from Supported to Contradicted with confidence. Use when someone asks "is this claim actually backed by the study it cites", "verify the claims in this whitepaper", "trace this statistic to its source", or needs an evidentiary audit of research, marketing, or policy claims. Do NOT use for a quick true/false corroboration of a discrete factual statement - use fact-checker instead; for broad open-ended investigation of a topic, use deep-research instead.
What it loads into context
Frontmatter
148 tokensAlways available to the agent
SKILL.md
1,541 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: filled verification
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytesSKILL.mdcore7,355 bytes
Read SKILL.md
# Claims Verifier
The job is to determine whether a claim is supported by evidence - not whether it sounds plausible. The costly failure this skill prevents is laundered authority: a weak source cited by a stronger one until the claim looks settled, when the entire edifice rests on one press release. Extraordinary claims require extraordinary evidence.
## Inputs to collect
1. **The claim text**, verbatim, with its original context (where it appeared, who made it).
2. **The stakes**: what decision the verdict will inform. High-stakes claims get deeper chain-tracing.
3. **Any citations already attached** to the claim.
4. **Access constraints**: paywalled sources the user can or cannot retrieve. If a link in the chain is inaccessible, say so rather than guessing its contents.
## Operating procedure
### Step 1: Decompose the claimclickhouse-analytics
When your agent uses it
Designs ClickHouse schemas and queries for fast analytics - MergeTree engine selection, ORDER BY key design, partitioning, materialized views, and projections. Use when someone asks "why is my ClickHouse query slow", "how should I order my sorting key", "should I use a materialized view or projection", "how do I deduplicate events", or is modeling an event or metrics table for OLAP. Do NOT use for tuning row-store OLTP databases like Postgres or MySQL - use sql-query-optimizer instead; for general relational schema design use database-schema; for reading query plans on traditional databases use explain-plan-reader; for the streaming ingestion side use kafka-pipelines.
What it loads into context
Frontmatter
157 tokensAlways available to the agent
SKILL.md
1,678 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Query rules
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,504 bytesSKILL.mdcore7,504 bytesSKILL.mdcore7,504 bytes
Read SKILL.md
# ClickHouse Analytics
ClickHouse is fast only when the schema matches the queries: columnar, sorted storage means the ORDER BY key does the work a row-store's indexes would. The expensive mistake is designing the table like Postgres - wrong sorting key, per-row inserts, SELECT * - and concluding ClickHouse is slow when it is scanning every granule because nothing let it skip.
## Operating procedure
Design in this order because the sorting key depends on the queries, the engine on the write semantics, and everything downstream (partitions, MVs, projections) on those two.
### Step 1: gather inputs
Collect before writing any DDL, labeling estimates as estimates:
- The top 5-10 queries by frequency and pain: which columns they filter on, group by, and over what time ranges.
- Ingest shape: rows/second, batch or streaming, and whether the source can send duplicates or updates.client-churn-save
When your agent uses it
Detect at-risk agency retainer clients early and run the save - a ranked early-warning signal list, the diagnose-before-prescribing save-call procedure with a word-for-word script skeleton, the re-scope-down-not-lose rule, and the 90-day post-loss win-back window. Use when an agency owner says "I think we're about to lose this client", "the client went quiet on us", "they said they're doing a budget review", "our main contact just left", or "we got a cancellation notice". Do NOT use for designing the routine communication rhythm that prevents churn in the first place - use client-comms-cadence instead.
What it loads into context
Frontmatter
140 tokensAlways available to the agent
SKILL.md
1,959 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore8,817 bytes
Read SKILL.md
# Client Churn Save
By the time a client says "we've decided to go another direction," the decision is weeks old and the save rate is near zero. The costly mistake this skill prevents is late detection followed by panic discounting: noticing churn only at the cancellation email, then throwing a price cut at a problem that was never about price. Churn saves are won in the detection window - the four to eight weeks between the first quiet signal and the internal decision - and won by diagnosis, not concessions.
Work the example agency throughout: Northbeam Digital, an 8-person marketing agency with 11 retainer clients averaging $6,500/month and a 62 percent gross margin target. Losing one average client costs $78,000 a year of revenue; at Northbeam's economics, replacing it takes a paid audit, a close, and onboarding - months of pipeline. A save is almost always cheaper than a replacement.
## Operating procedure
### Step 1: Watch the ranked early-warning signals
These are the escalation triggers from client-comms-cadence, ranked here by severity. The account manager escalates to the owner the same day any of them fires; two simultaneous signals mean the save call happens this week.
1. Champion departure - the day-to-day contact leaves or changes roles. Highest risk: the retainer's institutional memory and internal advocate leave together. Treat as an active save even if nothing else looks wrong.
2. "Budget review" language - "budget review," "procurement is looking at vendors," "re-evaluating for next year," or a competitor's name in any message. The internal conversation has already started.client-comms-cadence
When your agent uses it
Design an agency's client communication operating system - a channel contract per client tier, the weekly pulse / monthly report / quarterly QBR ladder, same-business-day response SLAs, the no-surprises rule for bad news, and internal escalation triggers. Use when an agency owner says "clients keep asking what we're working on", "every client pings us on a different channel at all hours", "we got fired and never saw it coming", or "my team is drowning in client Slack messages". Do NOT use for planning the content of the monthly report itself - use agency-monthly-report instead.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
1,874 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: channel contract template
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore8,585 bytes
Read SKILL.md
# Client Comms Cadence
Agencies rarely get fired for bad work; they get fired for silence. The client who "never hears from you" mentally cancels the retainer months before procurement does, and the costly mistake this skill prevents is reactive communication - answering whatever arrives, on whatever channel, whenever - which reads to the client as chaos and burns the team out. A communication cadence is a contract: the client knows exactly when they will hear from you, so they stop wondering whether the retainer is working.
Work the example agency throughout: Northbeam Digital, an 8-person marketing agency with 11 retainer clients averaging $6,500/month and a 62 percent gross margin target.
## Operating procedure
Set tiers first, then channels, then the ladder - the cadence a client gets is a function of what they pay, and per retainer-economics-calculator, account manager communication time is delivery cost, so over-communicating with a small retainer destroys its margin.
### Step 1: Tier the client book
Split clients into two or three tiers by retainer size and strategic value. Northbeam uses two: Tier A at $7,000/month and up (5 clients: Crestline SaaS, Verra Health, Pillar Home, Onyx Fitness, Bluepine Outdoors), Tier B below (6 clients). Tiering is internal; clients see their own contract, never the tier list.
client-onboarding-system
When your agent uses it
Runs a first-14-days client onboarding procedure - kickoff agenda, asset and access collection checklist, a written communication contract, and the first-quick-win rule of shipping something visible in week one. Use when a freelancer or consultant says "new client just signed, now what", "how do I kick off this engagement", "set up my client onboarding", "the project stalled waiting on client assets", "client keeps Slacking me at 10pm", or "how do I start a retainer on the right foot". Do NOT use for onboarding a new employee into a role - use onboarding-plan-builder instead; for in-product user onboarding copy, use onboarding-copy.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,887 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Concrete thresholds
- Deliverable
- Do NOT
- Quality bar
- Neighbors
Files
SKILL.mdcore8,642 bytesSKILL.mdcore8,642 bytes
Read SKILL.md
# Client Onboarding System
The first 14 days of an engagement set the operating norms for its entire life: who communicates where, how fast things move, and whether the client experiences the freelancer as a professional operation or an improvising individual. The costly mistake this skill prevents is the drift start - a signed contract followed by two weeks of scattered emails, missing assets, and no visible output, which trains the client to micromanage before any work has shipped.
Run this immediately after the SOW from statement-of-work-writer is signed and the deposit or first retainer invoice has cleared. The cadence established here is what client-status-reporting runs on from week 2 onward. Worked example: Maya, a freelance brand designer, onboarding her first Brand System Retainer client ($4,900/month, up to 36 hours) - a DTC hot-sauce brand.
## Operating procedure
### Step 1: Collect the inputs
1. The signed SOW: deliverables, cadence, and the designated approver.
2. Payment confirmation. No onboarding before the first invoice clears - starting work on promise-to-pay undoes the payment discipline set in statement-of-work-writer.
3. The client-side cast: approver, day-to-day contact, and anyone whose assets or access are needed.
4. A candidate first quick win (see Step 5) - pick it before kickoff so the first week has a target. Label it a guess until the client confirms priorities.client-status-reporting
When your agent uses it
Produces the weekly 5-block client status report - done, next, blocked, decisions needed, budget/hours - in a 15-minute time cap, from a copy-paste template with [FILL] fields. Use when a freelancer or consultant says "client asked what they're paying for", "how do I keep clients updated", "write my weekly status update", "client keeps pinging me for progress", "they went quiet and I'm worried about the renewal", or "how do I show retainer value". Do NOT use for long-form consulting or research reports - use formatted-report-writer instead; this skill is the recurring one-page operating rhythm of an engagement.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,861 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template: weekly status report
- Worked example: Maya's week-3 report
- Concrete thresholds
- Deliverable
- Do NOT
- Quality bar
- Neighbors
Files
SKILL.mdcore8,123 bytes
Read SKILL.md
# Client Status Reporting
Clients do not churn because the work was bad; they churn because they could not see the work at all. "What am I paying for?" is never really a question - it is the announcement that invisibility has already done its damage, and by then the renewal is in doubt. The weekly status report is the cheapest retention instrument a freelancer owns: fifteen minutes that preempts check-in pings, forces decisions out of clients on a schedule, and makes the invoice feel obvious instead of debatable.
In the Freelancer / Consultant OS pack, this skill takes over from client-onboarding-system at day 14 and runs for the life of the engagement. Worked example: Maya, a freelance brand designer on a $4,900/month Brand System Retainer (up to 36 hours) with a DTC hot-sauce brand.
## Operating procedure
### Step 1: Collect the inputs (once, at setup)
1. The communication contract from client-onboarding-system: channel, recipient(s), and send day. Default: same weekday every week (Friday morning or Monday morning), in the engagement's single async channel.
2. The SOW deliverables and, for retainers, the monthly hours cap - the budget block reports against these.
3. Where hours are tracked. If they are not, start now; the budget block cannot be guessed. Label any estimated figure as an estimate.
clinical-summary
When your agent uses it
Produces structured, faithful clinical summaries - either a patient-record summary with strict chronology, medication reconciliation, and expanded abbreviations, or a clinical-study summary with effect sizes, absolute risks, harms, and limitations. Use when someone asks "summarize this patient chart", "turn these notes into a discharge-style summary", "summarize this RCT for journal club", or "what did this trial actually show". This is clinical documentation support, not medical advice or diagnosis. Do NOT use for explaining lab results to a patient - use bloodwork-explainer instead; for synthesizing many studies into one evidence review, use systematic-review.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,524 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Path A: patient-record summary
- Path B: clinical-study summary
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,357 bytesSKILL.mdcore7,357 bytesSKILL.mdcore7,357 bytesSKILL.mdcore7,357 bytesSKILL.mdcore7,357 bytes
Read SKILL.md
# Clinical Summary
A clinical summary gets acted on by people who trust it, so the costly failure is not vagueness - it is confident distortion: a scrambled timeline, a dropped medication, an overstated trial result. Faithfulness to the source is paramount; never state more certainty than the documents support, and never fill a gap with a plausible guess. This skill produces two artifact types with one discipline: patient-record summaries and clinical-study summaries.
**Boundary, stated up front:** this is clinical documentation support - organizing and restating what the source material says. It is not medical advice, not diagnosis, and not treatment recommendation. Never suggest a diagnosis the records do not contain, never recommend starting, stopping, or dosing a medication, and always defer care decisions to the treating clinician.
## Inputs to collect
1. **Mode** - patient-record summary or study summary. If the source is a chart, notes, or discharge paperwork, it is Path A; if a paper or trial report, Path B.
2. **Audience** - clinician, care coordinator, reviewer, or the patient's own file. Default: clinician-facing.
3. **Source documents** - everything available. Note explicitly which document types are missing (no med list, no imaging reports), because gaps change the summary's reliability.
4. **Purpose** - handoff, chart review, journal club, appraisal. Purpose sets length and emphasis.
Label every inference as an inference ("dose change appears to follow the elevated creatinine on [date], though no note states the reason").closer-sales-script
When your agent uses it
Produce and run a word-for-word gym intro-consult script using the CLOSER sequence (Clarify, Label, Overview, Sell, Explain, Reinforce) to take a booked lead to a paid membership. Use when a gym owner or coach asks for a sales script, says "how do I close the consult", "intro consult", "no-sweat consult", "CLOSER framework", "present the price", "I'm bad at selling memberships", or wants to prep for or review an intro consult. Do NOT use when the request is dissolving a specific objection, speeding up lead response, or building a follow-up cadence - use objection-handling-and-speed-to-lead instead; do NOT use to set the price or guarantee - use gym-pricing-and-guarantees instead.
What it loads into context
Frontmatter
169 tokensAlways available to the agent
SKILL.md
2,149 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Deliverable
- Template: consult-script
- Template: discovery-questions
- Template: consult-tracker
Files
SKILL.mdcore8,960 bytes
Read SKILL.md
# CLOSER Sales Script
Run the booked intro consult as a structured conversation that ends in a paid membership: understand the prospect, show them the outcome, present price once desire exists, then lock the next step.
## Workflow
Run the six stages strictly in order, inside a 30-45 minute consult - long enough to build real desire, short enough that the decision happens in the room. Do not present price until Clarify, Label, Overview, and Sell are done - price before desire kills the sale.
1. **Clarify why they came.** Open warm, then get the goal and the why behind it. Ask: "What made you book today?", "What do you want to change in the next few months?", "Why now - what's different about this moment?" Write down their exact words; you reuse them in every later stage. The prospect should be doing most of the talking here - if you are speaking more than about a third of the time in the first three stages, you are pitching, not clarifying.
2. **Label the problem in their words.** Reflect it back until they confirm it: "So the real issue is [their words], is that right?" Get an explicit yes. Reflect their words; do not tell them what their problem is.
3. **Overview past attempts and their cost.** Surface why the old way failed and what staying the same costs: "What have you tried? How did it go? What has that cost you - not just money, but how you feel?" This builds urgency; never use it to shame.
cma-narrative-builder
When your agent uses it
Turns a comparative market analysis into a pricing story a seller accepts - defensible comp selection, plain-language adjustments, the overpricing-cost math, and a price-band recommendation aligned to portal search brackets. Use when an agent says "help me present my CMA", "the seller wants to list too high", "how do I justify this price", or "Zillow says their house is worth more". Do NOT use for negotiating the offers that come in after listing - use re-negotiation-prep instead.
What it loads into context
Frontmatter
112 tokensAlways available to the agent
SKILL.md
2,483 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Comp selection rules - and how to defend each comp
- Adjustments in plain language
- The overpricing-cost math
- The price band and the search-bracket rule
- Objection scripts
- Deliverable
- Do NOT
- Quality bar
- Escalation and compliance
Files
SKILL.mdcore10,845 bytesSKILL.mdcore10,845 bytes
Read SKILL.md
# CMA Narrative Builder
A CMA is not a spreadsheet problem, it is a persuasion problem: the data almost always supports a number lower than the seller's hope, and the listing dies or thrives on whether the agent can tell the pricing story convincingly enough that the seller prices to the market instead of to their memories. The costly mistake this skill prevents is taking an overpriced listing to keep the seller happy - the property sits, goes stale, and sells below what correct pricing would have fetched, and the agent wears the failure.
Worked example throughout: Priya, a residential agent in a suburban market, 14 transactions last year at a $485,000 average sale price, building toward 24. The subject property is 742 Alder Court - 4-bed, 2.5-bath, 2,340 sq ft colonial, 2022 kitchen renovation, new roof, 0.4 miles to commuter rail (the same worked listing as listing-description-writer). The sellers believe it is worth $525,000 because "Zillow says $521k." The comps support $495,000-$505,000.
## Operating procedure
1. **Select comps by the three rules before touching adjustments.** Selection is where CMAs are won or lost; a weak comp poisons every number downstream, so lock the comp set first. Aim for 3-5 solds plus 2-3 actives (the actives show the competition, not the value).
2. **Adjust each comp to the subject** using the plain-language adjustment method below. Adjust the comp toward the subject, never the reverse.
3. **Run the overpricing-cost math** so the "just try it high for a few weeks" conversation is answered with numbers before it is raised.
4. **Set the price band and snap to the search bracket** - $499,000, not $505,000.
5. **Build the presentation in this order:** comps and their stories first, adjustments second, the market-time math third, the recommendation last. Sellers who hear the number before the evidence argue with the number; sellers who see the evidence first often say the number themselves.
6. **Rehearse the objection scripts,** especially "Zillow says more."code-review-checklist
When your agent uses it
Run a systematic multi-pass code review - correctness, design, security, performance, tests - and report findings ordered by severity with concrete, respectful suggestions. Use when someone asks "review this PR", "review this diff", "what's wrong with this change", or wants a pre-merge quality gate on a branch. Do NOT use for a security-only deep audit of a change - use secure-code-review instead - or for writing the ticket or tracking artifact that describes the change - use jira-ticket-writer instead.
What it loads into context
Frontmatter
122 tokensAlways available to the agent
SKILL.md
1,337 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked contrast - phrasing a finding
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,265 bytes
Read SKILL.md
# Code Review Checklist
A good review catches the bug before production and leaves the author better at their job; a bad review rubber-stamps the diff or nitpicks formatting while an injection hole ships. This skill turns review into an ordered procedure - separate passes with separate questions - because reading once for everything at once is how correctness bugs hide behind style comments.
## Inputs to collect
Before the first pass, gather: the diff itself, the stated intent (PR description, ticket, or ask the author - reviewing a change without knowing what it claims to do is guessing), and how the change reaches users (user-facing, internal tool, data migration) since that sets the severity bar. Review the change, not the whole codebase: read surrounding context only where the diff demands it.
## Operating procedure
Run the passes in this order. Correctness comes first because a beautifully designed wrong answer is worthless, and security before performance because a fast vulnerability is worse than a slow safe path. Do each pass as a separate read of the diff.
### Pass 1 - Correctness
cold-email-craft
When your agent uses it
Write short, personalized B2B cold emails and follow-up copy - under 90 words, one ask, personalization in the first line - with good/bad contrast pairs to calibrate. Use when someone asks "write a cold email", "why is nobody replying to my outreach", "draft a follow-up sequence", "make this email shorter", or is starting founder-led outbound. Do NOT use for sending infrastructure, domains, SPF/DKIM/DMARC, warmup, or spam placement - use cold-email-deliverability instead; for multi-channel cadence architecture use outreach-sequence-designer; for building and verifying the list use prospect-list-builder.
What it loads into context
Frontmatter
151 tokensAlways available to the agent
SKILL.md
1,600 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked pair
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,926 bytes
Read SKILL.md
# Cold Email Craft
A cold email earns a reply by respecting the reader's time and making the next step trivial. The costly mistake this skill prevents is the 200-word "quick question" pitch that reads as mail-merge, burns the prospect, and teaches the founder that outbound doesn't work. Short beats clever; this skill owns the words. If the words are good but nothing lands in the inbox, the problem is infrastructure - route to cold-email-deliverability.
## Operating procedure
### Step 1: Gather inputs
1. Who is being emailed: role, company profile, and one concrete reason this person, now (a launch, a hire, a post, a funding event). If no reason exists, the list is not ready - route to prospect-list-builder.
2. The offer and the one result it produces, with a number from a comparable company. Default: if no customer result exists yet, use the founder's own before/after and label it as such.
3. The lowest-friction ask the sender can honor ("15-min call", "2-min Loom", "reply with yes/no").
4. Sequence length and channel mix already planned, if any (architecture belongs to outreach-sequence-designer; copy for each step belongs here).
### Step 2: Write the first line - personalization firstcold-email-deliverability
When your agent uses it
Sets up and protects cold-outbound sending infrastructure - dedicated lookalike domains, SPF/DKIM/DMARC authentication, 2-4 week mailbox warm-up, 30-50 sends/day per-mailbox caps, and continuous bounce/complaint monitoring, with a runnable DNS audit script. Use when someone says "my emails are going to spam", "set up cold email infrastructure", "warm up a new domain", "configure SPF DKIM DMARC", "how many cold emails can I send per day", or "my domain got blacklisted". Do NOT use for protecting an opted-in marketing or transactional program on your primary domain - use email-deliverability instead; do NOT use for writing the email copy - use cold-email-craft instead; do NOT use for designing the multi-step cadence - use outreach-sequence-designer instead; do NOT use for sizing the list - use prospect-list-builder instead.
What it loads into context
Frontmatter
206 tokensAlways available to the agent
SKILL.md
4,149 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- The workflow
- Pre-flight checklist
- Worked example: infrastructure for 200 cold sends/day
- Authentication audit script (Node, zero deps)
- Deliverable
- Common failure modes
- Quality bar
Files
SKILL.mdcore16,242 bytesSKILL.mdcore16,242 bytesSKILL.mdcore16,242 bytesSKILL.mdcore16,242 bytesSKILL.mdcore16,242 bytesSKILL.mdcore16,242 bytes
Read SKILL.md
# Cold Email Deliverability
Deliverability is not a copywriting problem; it is an infrastructure and reputation problem. The single most expensive mistake in cold outbound is sending it from the primary corporate domain - one spam-trap hit or complaint spike and invoices, hiring emails, and the founder's replies all start landing in spam, and that damage is slow and painful to reverse. Treat cold sending as a blast radius to isolate on purpose: separate domains, separate mailboxes, conservative volume, relentless monitoring.
Mailbox providers (Gmail, Outlook) score the *sending identity* - domain + IP + behavioral signals like complaints, bounces, and engagement - not the cleverness of the subject line. Primary-inbox placement is earned by behaving like a real human correspondent at low volume. Good copy on a burned domain still lands in spam. Note the scope: cold outreach to people who never opted in. An opted-in newsletter or transactional program follows different rules (subdomains of the primary domain, higher warm-up ramps, sunset policies) - that is email-deliverability territory.
## Inputs to collect
- Target daily send volume (or monthly prospect goal ÷ ~20 business days).
- Primary corporate domain, and confirmation that it will NOT be used for cold sends.
- Existing sending domains, their age, and any prior reputation damage (blacklist history, complaint spikes).
- ESP/mailbox provider (Google Workspace, Microsoft 365) and any existing SPF includes on the sending domains.
- List source and whether addresses are verified. If unverified, stop - route through lead-enrichment first.
- Label any volume or list-quality figure the user estimates rather than measures as a guess.color-accessibility
When your agent uses it
Audits and repairs color palettes against WCAG contrast thresholds - 4.5:1 for body text, 3:1 for large text and UI components, 7:1 for AAA - and makes them safe for color-blind users, delivering a verified pairing table. Use when someone asks "does this color pass contrast", "make my brand color accessible", "is this palette color-blind safe", "why did my contrast check fail", or is remediating color findings from an audit. Do NOT use to invent a brand palette from scratch - use color-palette-builder instead; for a full accessibility review beyond color, use accessibility-audit.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
2,133 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Contrast calculator
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,448 bytesSKILL.mdcore8,448 bytesSKILL.mdcore8,448 bytes
Read SKILL.md
# Color Accessibility
A palette that fails contrast ships a product that a measurable slice of users cannot read: low-vision users lose the text, and roughly 8% of men (and about 0.5% of women) with red-green color blindness lose any meaning carried by hue alone. The costly mistake this skill prevents is approving colors that look fine on a designer's bright monitor and fail on real screens, in sunlight, and in audits - forcing a token-system rework after components already reference the broken values.
## Operating procedure
Work the steps in order: pairings must be enumerated before ratios are computed, and ratios must pass before color-blindness simulation, because a palette that fails contrast is not worth simulating.
### Step 1: Gather inputs
Collect before starting. Label any assumed value as a guess.
1. Brand and functional colors as hex values (primary, semantic red/green/yellow, neutrals).
2. Every background surface they sit on - including dark mode surfaces if one exists.color-palette-builder
When your agent uses it
Builds an accessible brand color palette - one anchor color, role assignments, 9-step tonal scales, 60-30-10 distribution, and documented usage rules - ready for a developer to implement without guessing. Use when someone asks "pick our brand colors", "build a color system for our app", "what shade should our buttons be", or is defining the color layer of a design system. Do NOT use for auditing or fixing contrast failures and color-blindness issues in an existing product - use color-accessibility instead; for assembling the full brand book around the palette, use brand-guidelines.
What it loads into context
Frontmatter
140 tokensAlways available to the agent
SKILL.md
1,663 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,513 bytes
Read SKILL.md
# Color Palette Builder
A brand color palette is not a mood board - it is a functional system with roles, scales, and rules. The costly failure is shipping pretty swatches with no usage rules: developers guess, screens drift, and within months the product has fourteen slightly different blues and body text that fails contrast. Build the palette so a developer can implement it without asking a single question.
## Operating procedure
Work the steps in order: roles before scales, scales before contrast checks - contrast is verified per color pair, and pairs only exist once the scales do.
### Step 1: Gather inputs
Collect these before generating anything; label any guess as a guess.
- Anchor color, if one exists (hex value). If none, collect 3-5 brand adjectives plus the category's color conventions to derive one.
- Surfaces in scope: marketing site, product UI, or both. Product UI needs a deeper neutral scale.community-engagement-playbook
When your agent uses it
Builds the student community system for a paid course - platform choice by where students already are, an introduce-yourself onboarding ritual, a weekly cadence of office hours, wins threads, and accountability checks, a hard creator-time budget, and scheduled testimonial-harvest moments. Use when a course creator says "my students aren't finishing the course", "should I use Circle or Slack or Discord", "how do I keep my community active without it eating my week", or "how do I get testimonials from students". Do NOT use for running the testimonial conversation itself - use testimonial-capture-interview instead.
What it loads into context
Frontmatter
128 tokensAlways available to the agent
SKILL.md
1,984 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore9,301 bytes
Read SKILL.md
# Community Engagement Playbook
A course community exists for exactly two outputs: completion and testimonials. Students who feel seen finish, finished students get results, and results become the proof that sells the next cohort. The costly mistake is treating community as a feature to bolt on - an empty Slack nobody posts in signals a dead course louder than no community at all, and an unbudgeted community quietly consumes the creator's production time until the next launch slips.
This skill runs after launch-email-sequence hands over the buyers, operates against the completion target set by course-outline-architect, and feeds its testimonial harvest back to course-sales-page's proof section. That loop - community to testimonials to sales page - is what makes the second launch outperform the first.
Worked example throughout: Dana, who just closed the "Notion OS for Consultants" launch ($297) with roughly 120 students from her 8,000-subscriber list, budgeting 5 hours a week for community.
## Operating procedure
### Step 1: Collect the inputs
1. Student count and where the audience already hangs out (check where they replied from during the launch; ask 5 buyers directly if unknown).
2. The course's quick-win moment and module milestones from course-outline-architect.comparison-page-builder
When your agent uses it
Builds an honest "X vs Y" or "best alternative to X" comparison page - scannable table plus use-case routing copy - for high-intent commercial searches. Use when someone asks "write a page comparing us to a competitor", "build an X vs Y comparison page", "create an alternative-to page", or wants a competitor comparison table and copy that names a rival and helps the reader choose. Do NOT use for a general marketing landing page - use landing-page-copy instead; for competitor research itself, use competitive-intelligence.
What it loads into context
Frontmatter
127 tokensAlways available to the agent
SKILL.md
1,055 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,856 bytesSKILL.mdcore4,856 bytesSKILL.mdcore4,856 bytesSKILL.mdcore4,856 bytes
Read SKILL.md
# Comparison Page Builder
Build a comparison page that wins "X vs Y" and "best alternative to X" searches by being genuinely fair, because a shopper who catches one false claim about a rival distrusts every claim about you.
## Workflow
1. Gather inputs: your product, the named competitor(s), the buyer segment you serve best, and the real, verifiable differences. If the requester cannot supply verifiable differences, stop and ask - do not invent them.
2. Concede at least one real point: name something the competitor genuinely does better. Honest concession is what makes every other claim believable; "we win on everything" reads as marketing and converts worse.
3. Build the table as the page centerpiece, placed within the first scroll. Pick 6-10 criteria that drive the buying decision (price, key features, support, warranty, ease of use, fit for use case), not vanity specs. One short value, phrase, or checkmark per cell. Attribute and date competitor figures; where a value is unknown, write "varies" or omit the cell - never fabricate a number.
4. Frame by use case: add "Choose us if…" and "Choose them if…" blocks above or below the table. This routes the right buyer in, the wrong-fit buyer out (fewer refunds, better retention), and captures "X vs Y for [use case]" long-tail intent.
5. Optimize for the query: use natural "X vs Y" phrasing in the H1, title, and intro; answer the implicit "which is better" question in the first paragraph - a direct 40-60 word answer is the format featured snippets favor; add an FAQ of 4-6 "Is X or Y better for…" questions; link to the relevant product page with descriptive anchor text.
6. Run the legal/ethics check before shipping: every comparative claim is substantiated, accurate, and dated; no disparagement, no misleading trademark use, no stale pricing. Put a re-verification date on the calendar - competitor pricing and plans should be re-checked at least quarterly, and a table whose sources are more than 6 months old is a liability, not an asset.
## Deliverable
Produce a publish-ready comparison page containing:competitive-intelligence
When your agent uses it
Map competitors across positioning, product, pricing, proof, and weaknesses, then turn the research into sales-ready battlecards with an update cadence. Use when someone asks "what are our competitors doing", "build a battlecard", "how do we compare to X", "why are we losing deals to X", or before a pricing or positioning decision. Do NOT use for designing long-term defensibility strategy - use competitive-moat instead; for a public comparison landing page use comparison-page-builder; for rep-facing objection scripts beyond the battlecard use objection-handler.
What it loads into context
Frontmatter
128 tokensAlways available to the agent
SKILL.md
1,522 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Battlecard template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytesSKILL.mdcore6,756 bytes
Read SKILL.md
# Competitive Intelligence
Competitive intelligence turns scattered impressions of rivals into artifacts sales and product can act on. The costly mistake it prevents is losing winnable deals to a competitor pitch nobody prepared for - or worse, repositioning the whole product around a competitor claim that was never verified. Everything below runs on one discipline: separate fact from inference, and date every claim.
## Operating procedure
### Step 1: Gather inputs
1. The 3-5 competitors that actually show up in deals (ask sales; if pre-sales, the ones prospects mention). Default: cap at 3 - depth beats coverage.
2. Which deals or segments each appears in.
3. Existing win/loss notes, lost-deal reasons, sales call recordings if available.
4. The decision this research feeds: sales enablement, pricing move, roadmap, or positioning. This sets how deep each dimension goes.
### Step 2: Work the source ladder, primary firstcompetitive-moat
When your agent uses it
Identifies which of the seven structural moats a business actually has, pressure-tests each claim against funded copycats and incumbent pivots, and delivers a moat scorecard with one moat to invest in deliberately. Use when someone asks "what's our moat", "is our product defensible", "a well-funded competitor just launched - what protects us", "investors keep asking about defensibility", or "which moat should we build first". Do NOT use for tracking what competitors are shipping and saying - use competitive-intelligence instead - or for finding unserved market segments - use white-space-analysis instead.
What it loads into context
Frontmatter
139 tokensAlways available to the agent
SKILL.md
1,612 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: scorecard contrast
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,677 bytes
Read SKILL.md
# Competitive Moat Analysis
A moat is a structural reason an advantage compounds rather than erodes. Most "moats" founders cite - features, a head start, proprietary data - are things a funded competitor copies in a quarter, and the costly mistake is building a fundraise or a strategy on one of them. This skill separates real moats from wishful thinking and names the one moat worth building deliberately.
## Operating procedure
Run taxonomy before testing, and testing before measurement - you cannot stress-test a moat you have not precisely named, and you should not bother measuring one that fails the adversary test.
### Step 1: Gather inputs
1. The claimed advantages - everything the founder or team believes protects them, verbatim.
2. Business model and stage - who pays, how much, how customers onboard and leave.
3. Competitive set - the funded startups and the incumbents, with rough war chests.
4. Retention data if available - gross retention by cohort age; label estimates as guesses.component-api-design
When your agent uses it
Design clean, composable React/Vue component APIs with accessibility built in. Use when designing or reviewing a component's props, composition, events or accessibility, or when a component API has grown hard to use.
What it loads into context
Frontmatter
60 tokensAlways available to the agent
SKILL.md
659 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Principles
- Step 1: Name the responsibilities
- Step 2: Choose the composition model
- Step 3: Design the props
- Step 4: Events
- Step 5: Accessibility contract
- Step 6: Write the usage examples first
- Output
Files
SKILL.mdcore2,930 bytes
Read SKILL.md
# Component API Design
Use this skill to design the public API of a UI component before writing
implementation. A good API is learnable, hard to misuse, and composable.
## Principles
1. **Make the common case effortless, the complex case possible.**
2. **Prefer composition over configuration** - slots beat boolean explosions.
3. **The prop name is documentation** - it should read aloud as English.
4. **Accessibility is not a prop** - it is the default behavior.
## Step 1: Name the responsibilities
conflict-resolution
When your agent uses it
Facilitates workplace and interpersonal conflict resolution using Nonviolent Communication and interest-based negotiation, producing a mediation plan, de-escalation scripts, and a written agreement with a check-in date. Use when someone asks "how do I resolve a conflict with my coworker", "two of my reports are fighting", "help me have a hard conversation", or "how do I de-escalate an argument". Do NOT use for de-escalating angry customers demanding refunds - use refund-deescalation instead; for writing a team's standing conflict norms into a charter, use team-charter.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
1,374 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,513 bytes
Read SKILL.md
# Conflict Resolution
Most conflicts stall because people negotiate positions ("I want X") instead of interests ("I want X because I need Y"). Positions collide; interests usually overlap. The costly mistake this skill prevents is splitting the difference on positions - a compromise that leaves both parties resentful - when a solution meeting both underlying interests was available the whole time.
## Operating procedure
Order matters: understanding must come before solutions, because solutions imposed on people who do not yet feel heard get rejected regardless of merit.
### Step 1: Gather inputs
Collect before advising. Label anything inferred as a guess.
1. The parties and the relationship (peer, manager-report, cofounders, family).
2. The presenting position of each side, in their own words.connection-pool-tuner
When your agent uses it
Diagnoses connection pool exhaustion from captured evidence and sets correct pool sizes and timeouts across app pools and PgBouncer. Use when someone says "we keep hitting database connection limit errors under load", "my API times out waiting for a connection from the pool", "the database shows hundreds of idle connections", or sees "too many clients", checkout/acquire timeouts, "idle in transaction" pile-up, or serverless functions storming the database on a traffic spike. Do NOT use when the problem is an unreliable external service (third-party API, downstream microservice) needing retries or fallbacks - use circuit-breaker-builder instead.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,246 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
- Defaults to reach for
Files
SKILL.mdcore5,681 bytes
Read SKILL.md
# Connection Pool Tuner
Pool exhaustion shows up as checkout timeouts and "too many clients" errors; the fix is almost never "raise max_connections." Diagnose from evidence first, then size to the database.
## Workflow
1. **Capture the evidence before touching any number.** Query `pg_stat_activity` grouped by `state` under load. Record total connections, count per state, and the longest-running query/transaction age. Also record every pool that points at the DB: app replicas times per-replica pool size, plus migrations and any superuser sessions.
2. **Classify the failure from the evidence:**
- Many `idle in transaction` rows -> code opens a transaction and does slow work (an HTTP call, sleep, batch) before commit. Fix the transaction scope, not the pool.
- Many `active` long-runners -> slow queries saturating connections. Tune the queries.
- Checkout/acquire timeouts while DB connections sit mostly idle -> the app pool is too small or connections leak (not returned to the pool). Find the leak before enlarging.
- Total connections at or above `max_connections` -> the sum of pools is too large; reduce per-replica pool or front with a proxy. Do not raise `max_connections`.
3. **Size the pool to the database, not the app.** Postgres performs best at a small connection count; each backend is a process with real memory cost. Target tens of active connections, not thousands; a common ceiling is roughly `(core_count * 2) + effective_spindle_count`. The sum of all pools must stay under `max_connections` with headroom for migrations and superuser - keep at least 10-15% of `max_connections` free. Twenty replicas with a pool of 20 each is 400 connections - usually far too many.content-brief
When your agent uses it
Builds an SEO content brief a writer can execute without guesswork - primary and semantic keywords, search intent, recommended format matched to the SERP, title, H1, meta description, a full H2/H3 outline, word-count guidance derived from ranking pages, internal links with anchors, featured-snippet opportunities, CTA, and the content gap competitors miss. Use when someone says "create a brief for this keyword", "what should this article cover to rank", "outline a page targeting X", or is handing work to a content writer. Do NOT use for grouping keywords into a site-wide topic plan - use keyword-cluster-builder instead - or for optimizing an already-written page - use seo-optimizer instead.
What it loads into context
Frontmatter
160 tokensAlways available to the agent
SKILL.md
789 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- Brief components
- Determining intent
- Outline quality
- Word count guidance
- Anti-patterns
- Output
Files
SKILL.mdcore3,609 bytes
Read SKILL.md
# Content Brief
You create the brief a writer follows to produce a page that ranks in search and serves the reader. The brief removes guesswork.
## Process
1. Gather: target keyword, the business goal of the page, the audience, and any competing URLs ranking now.
2. Determine search intent (below) - everything flows from it.
3. Build the brief.
## Brief components
1. **Primary keyword** - the exact phrase to target.
2. **Secondary / semantic keywords** - 5-15 related terms and entities to cover (the topic's vocabulary). These help relevance, not just repetition.content-refresh-auditor
When your agent uses it
Produces a prioritized refresh plan for a decaying published page - a refresh/consolidate/retire verdict plus ordered update actions to recover or grow rankings. Use when an existing URL is losing traffic or rankings, during a periodic content-decay sweep, or when asked to "update", "refresh", or "revive" an old article. Do NOT use when planning coverage for a draft that is not yet published or ranking - use SERP Gap Analyzer instead.
What it loads into context
Frontmatter
114 tokensAlways available to the agent
SKILL.md
1,086 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,144 bytes
Read SKILL.md
# Content Refresh Auditor
Most ranking decay is the SERP moving on while a page stood still: intent shifted, competitors expanded, facts aged. Refreshing a page that already holds authority and links is usually higher ROI than writing a new one. Produce a concrete diff between what the page says and what the query now demands, then a prioritized action list.
## Workflow
1. **Gather evidence first.** Pull the page's current position, traffic trend, and backlinks, and load the live top 10 for its target query. A sustained decline of 20-30% year over year is the usual trigger for a page entering the audit; a week-over-week wobble is not. Do not recommend any action before you have read today's SERP and the page's current standing - refresh decisions are evidence-gated, not from memory.
2. **Decide refresh, consolidate, or retire.** Refresh pages with residual rankings (positions ~5-20), backlinks, or leftover traffic - they have equity to recover. Consolidate when two pages compete for one intent. Retire and 301-redirect thin pages with no referring domains and negligible traffic (under ~10 organic visits a month with no path to relevance is a defensible retire line). State the verdict explicitly before listing actions; never refresh a page that should be merged.
3. **Re-check intent against today's SERP.** Compare the page's format to what now ranks. If the SERP shifted from guides to comparisons, or now shows shopping or video packs, the page type may be obsolete. Realigning page type to current intent is often the single biggest lever.
4. **Diff against the page's lost coverage.** List subtopics, entities, tools, and People Also Ask questions the current top results cover that this page once won or now omits. Frame this as the delta the page has fallen behind on - not a from-scratch coverage map for a new draft (that is SERP Gap Analyzer's job).
5. **Update freshness signals honestly.** Replace outdated stats, years, screenshots, prices, and dead examples. Rewrite the title and meta if they reference an old framing. Fix broken links and prune outbound links to dead pages. Bump the modified date only when content genuinely changed - fake date bumps are detected and ignored.
6. **Rebuild internal links and answer blocks.** Add links from newer related pages into the refreshed article and update its own links to current targets. Restructure the lead and key sections into concise answer blocks to recapture featured snippets and AI citations that drifted to competitors.
7. **Prioritize by recovery potential.** Order every action by expected traffic recovery against effort. Lead with intent realignment and coverage gaps on the highest-equity pages. Set expectations on timing: movement typically registers 4-12 weeks after the recrawl, so schedule the re-measurement then, and run the decay sweep itself on a quarterly cadence rather than waiting for a traffic crisis.
## Deliverablecontent-repurposing
When your agent uses it
Fans one long-form pillar asset (video, podcast episode, or article) out into 10+ platform-native atomic pieces with a derivation map - clips, quote cards, threads, LinkedIn posts, newsletter and blog sections - each recast in the target platform's register and sequenced over the week. Use when someone says "turn this video into clips", "repurpose this podcast episode", "get more mileage out of this article", or "build a distribution plan from one piece of content". Do NOT use to re-express one finished post as the native equivalent of the same asset on each channel - use cross-platform-reformatter instead; do NOT use to mechanically crop, caption, and export a finished video for social feeds - use social-video-formatter instead; do NOT use to design the ongoing pillar-and-cadence publishing system itself - use creator-content-calendar instead.
What it loads into context
Frontmatter
197 tokensAlways available to the agent
SKILL.md
1,901 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked derivation map
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytesSKILL.mdcore8,409 bytes
Read SKILL.md
# Content Repurposing
One strong pillar piece contains a week of distribution, and most creators throw that away by publishing once and moving on. Repurposing is not copying and pasting: it is extracting the strongest signal from a long-form asset and recasting it in the native language of each platform. The costly mistake this skill prevents is the opposite failure - spraying weak, resized fragments across every channel, which burns audience trust faster than posting nothing.
## Operating procedure
Run the steps in order. The source audit must come before any derivative is written, because a derivative built on a weak moment fails no matter how well it is formatted.
### Step 1: gather inputs
Collect these before writing anything. If a value is inferred rather than stated, label it a guess.
1. The source asset: a link, transcript, or draft. Required - never repurpose from a summary of the source.
2. The platforms the creator actively maintains. Default set: their primary short-form video platform, X or LinkedIn (whichever matches the audience), and email if a list exists. Do not generate pieces for channels the creator has abandoned.contract-test-writer
When your agent uses it
Writes consumer-driven contract tests at service and API boundaries - Pact consumer tests, provider verification, broker publishing, and can-i-deploy gates - so an incompatible change fails a build instead of breaking integrations in production. Use when someone asks "how do I stop the backend breaking my client", "set up Pact between these services", "a renamed field broke a consumer that unit tests passed", or two separately deployed services must stay in sync. Do NOT use for designing a new API contract from scratch or reviewing its design - use api-design instead; do NOT use for end-to-end user-journey tests - use e2e-scenario-author instead; do NOT use for calls inside a single deployable - write a unit test.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,389 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,627 bytes
Read SKILL.md
# Contract Test Writer
Pin the shape of a network boundary from the consumer's side so a provider change that breaks any deployed consumer fails the provider's build, not production. The costly failure this prevents is the silent break: both teams' unit tests pass, both deploys go green, and the integration dies at runtime because a field was renamed or a status code changed and nothing in either pipeline knew the other side existed.
## Operating procedure
The consumer-driven order is the point: the consumer declares what it needs first, and the provider verifies against that declaration. Reversing it (provider publishes a schema, consumers hope) is how unread fields calcify and real dependencies go untracked.
### Step 1: Gather inputs
1. The boundary: consumer name, provider name, and the calls in play. Confirm it is a real network boundary you do not control end-to-end (cross-service call, client-to-API). For an internal function call inside one deployable, stop - write a unit test instead.
2. The fields the consumer actually reads from each response - from the consumer's code, not the provider's docs.
3. Broker availability: Pact Broker or PactFlow URL and write credentials (default: recommend standing one up; pacts committed to a repo go stale silently).
4. The environments where consumer versions run (prod, staging), for can-i-deploy.core-four-lead-engine
When your agent uses it
Pick the lead channels that fit a gym's stage, back daily activity out of a member goal, and build a 30-day lead plan across the Core Four (warm outreach, content, cold outreach, paid ads). Use when a gym owner says "I need more leads", "how do I get members", "what marketing should I do", "build a lead plan", "warm/cold outreach", or "what's a lead magnet". Do NOT use when the owner wants other people to bring in members (referrals, affiliate partners, agencies, employees) - use referral-and-affiliate-system instead; for paid-ad creative and budget depth, hand off to gym-meta-ads-funnel.
What it loads into context
Frontmatter
151 tokensAlways available to the agent
SKILL.md
2,342 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: 30-day-lead-plan
- references/core-four-grid
- references/lead-magnet-design
- references/volume-math
Files
SKILL.mdcore9,768 bytesSKILL.mdcore9,768 bytesSKILL.mdcore9,768 bytesSKILL.mdcore9,768 bytes
Read SKILL.md
# Core Four Lead Engine
Turn "get more members" into a measured 30-day plan with daily activity targets across the four ways a gym can reach people itself.
There are only four ways to let people know about the gym, mapped by who you contact (knows you, or stranger) and how many at once (one, or many): warm outreach (knows you, one-to-one), content (follows you, one-to-many), cold outreach (strangers, one-to-one), paid ads (strangers, one-to-many). Every tactic is a version of one of these.
## Workflow
1. **Pick the starting channel by audience and budget.** Start where the gym has an unfair advantage. Existing audience (email list, social following, past members): start with warm outreach and content - free, fast, highest conversion. Money but no audience: start with paid ads against a front-end offer - buys reach now. Neither: start with cold outreach and content - costs time, not money, and builds the audience the other channels need. Pick one or two channels the owner can run daily without fail; add more later.
2. **Set the volume math.** Back daily activity out of the member goal. The funnel runs reach to lead to booked to show to close. Decide the goal and the conversion rates, then run `lead_volume.js` to compute leads needed and daily activity per channel. See `references/volume-math` for estimating each rate. The point is to make the goal concrete: "8 warm conversations and 21 cold outreaches a day" is a plan; "get more members" is not.
3. **Run each channel with a protected daily minimum.** Each channel works only with consistent volume. Warm outreach: message people who know you with a genuine check-in, then a soft offer - minimum is a fixed number of conversations a day. Content: post on a fixed cadence, each piece showing a problem the gym solves and a transformation - minimum is a cadence you never miss. Cold outreach: reach local strangers with a relevant opener and a free lead magnet - minimum is a fixed number of new outreaches a day. Paid ads: a daily budget against the front-end offer - hand off to gym-meta-ads-funnel for creative and budget.
course-outline-architect
When your agent uses it
Designs a paid online course structure backward from the student transformation - before/after states, one module per milestone, 5-12 minute lessons with one action each, and completion-first sequencing that front-loads quick wins. Use when a creator says "help me outline my course", "how many modules should my course have", "my course feels bloated, what do I cut", or "how do I structure the course so people actually finish". Do NOT use for planning a single lesson in depth - use lesson-plan-builder instead - or for sequencing objectives across multiple courses - use curriculum-mapper instead.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
1,998 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: Dana's outline
- Template: outline skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore8,853 bytes
Read SKILL.md
# Course Outline Architect
Most paid courses are built forward from everything the creator knows, and the result is a 12-hour brain dump that students abandon by module two. Industry completion rates run a brutal 10 to 15 percent, and a course nobody finishes produces no results, no testimonials, and no word of mouth - which quietly kills every future launch. This skill designs the course backward from one transformation, so every module is a milestone and every lesson is one action, and it enforces the scope cuts that make finishing possible.
Install this skill first in the Course Creator Launch System pack. The transformation statement it produces is the raw material for course-sales-page and launch-email-sequence, and the completion targets it sets are what community-engagement-playbook exists to hit.
Work the example throughout: Dana, a productivity consultant with an 8,000-subscriber email list, building "Notion OS for Consultants" at $297.
## Operating procedure
Follow the steps in order. The transformation comes first because every later decision - module count, lesson scope, what to cut - is judged against it. Deciding modules before the transformation is how brain dumps happen.
### Step 1: Collect the inputs
course-sales-page
When your agent uses it
Writes the long-form sales page for a paid online course - above-the-fold promise, PAS-to-proof-to-program-to-price section order, value-stack price anchoring, an action-based guarantee, and an FAQ mapped to real objections. Use when a creator says "write my course sales page", "why isn't my course page converting", "how do I present the price", or "what goes in the FAQ". Do NOT use for a general product or SaaS landing page - use landing-page-copy instead - or for the emails that drive traffic to this page - use launch-email-sequence instead.
What it loads into context
Frontmatter
127 tokensAlways available to the agent
SKILL.md
2,104 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore9,300 bytes
Read SKILL.md
# Course Sales Page
A course sales page fails one way above all others: it sells the course instead of the transformation - modules, hours of video, and features, when the buyer only cares what their life looks like after. This skill produces the long-form page in the section order that converts cold-to-warm readers, anchors the price with a value stack, and answers objections before the buyer bounces to "maybe later" - which for course launches means never.
This skill runs after course-outline-architect, which supplies the transformation statement and the cut-list bonuses, and alongside launch-email-sequence, which drives the traffic and must quote this page's promise, stack, and guarantee word-for-word. Testimonials arrive over time from community-engagement-playbook; the page ships with what proof exists today and gets stronger every cohort.
Worked example throughout: Dana, a productivity consultant with an 8,000-subscriber list, selling "Notion OS for Consultants" at $297.
## Operating procedure
### Step 1: Collect the inputs
1. The transformation statement from course-outline-architect. If it does not exist, stop and build the outline first - a sales page for an undefined transformation is decoration.
2. The module list with milestone names, and the cut-list items destined for the bonus vault.cover-letter-writer
When your agent uses it
Writes a tailored cover letter from a resume and a job posting - mirrors the posting's top three requirements with resume evidence, opens with a company-specific hook, and stays under 300 words. Use when a user says "write me a cover letter for this job", "tailor my cover letter to this posting", "here's my resume and the job ad, draft a cover letter", or "why is my cover letter not getting responses". Do NOT use for writing or improving the resume itself - use resume-writer instead - and do NOT use for overall application strategy, targeting, or tracking - use job-application instead.
What it loads into context
Frontmatter
135 tokensAlways available to the agent
SKILL.md
1,689 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Concrete thresholds
- Worked contrast - opening line
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore8,061 bytes
Read SKILL.md
# Cover Letter Writer
A cover letter has one job - earn a resume read from a screener who spends under thirty seconds on it - and most letters fail in the first sentence by opening with "I am writing to apply for" and then restating the resume the screener is about to read anyway. The costly mistake this skill prevents is the generic letter - one that could be sent to any company with a find-and-replace - because screeners detect those instantly and treat them as evidence of low interest. A tailored letter answers three specific questions the posting is asking and proves the candidate did homework the other applicants skipped.
## Operating procedure
Follow these steps in order. Requirement extraction comes before any writing, because the letter's entire structure hangs on which three requirements it answers - draft prose first and the letter drifts back into a resume summary.
1. **Collect the inputs.** Get the resume, the full job posting, and any company hook material (see Inputs below). Do not proceed on the posting title alone; the ranked requirements live in the body text.
2. **Extract the three requirements the posting cares most about.** Rank every stated requirement by three signals - repetition (appears in the summary and again in the bullet list), position (first three bullets outrank the last three), and specificity (a named tool, quota, or scale beats a soft trait like "team player"). Take the top three. Exactly three: fewer looks thin, more dilutes each into a mention instead of an argument.
3. **Mirror each requirement with resume evidence.** For each of the three, find the single strongest proof in the resume - an accomplishment with a number wherever one exists (revenue, users, percent improvement, team size, time saved). Use the posting's own vocabulary in the mirror sentence so a keyword-scanning reader ties evidence to requirement without effort. If the resume has no evidence for a requirement, say so to the user and pick the nearest adjacent proof - do not fabricate.
4. **Write the company-specific hook as the first line.** One sentence that could only be written to this company - a recent launch, a stated mission the candidate genuinely connects to, a product they actually use, a public challenge the role clearly exists to solve. If no hook material was provided, derive one from the posting itself and label it a guess for the user to confirm or replace.
5. **Assemble the letter.** Structure - hook line, one bridge sentence naming the role, one short paragraph (or tight two-sentence block) per mirrored requirement, and a closing line with a plain ask for a conversation. No "To Whom It May Concern"; use the hiring manager's name if known, otherwise the team name.
6. **Cut to under 300 words.** Target 150 to 250. Cut throat-clearing ("I believe that", "I am confident"), any sentence that restates a resume line without adding context, and all adjectives about the candidate's own character ("passionate", "driven") - evidence carries the claim or nothing does.coverage-gap-finder
When your agent uses it
Produces a risk-ranked list of untested critical paths and branches from a real branch-coverage report crossed with git churn, naming the specific missing cases and the smallest test that buys the most safety. Use when someone says "we have 85% coverage but I don't trust it", "what should we test next", "audit the suite before this release", or "which coverage gaps actually matter". Do NOT use to prove existing assertions are strong - covered-but-unasserted lines are mutation-test-runner's job; do NOT use to pin the current behavior of legacy code before a refactor - use characterization-test-writer instead. This skill decides what to test and in what order, not how well a test asserts or how to lock down legacy.
What it loads into context
Frontmatter
170 tokensAlways available to the agent
SKILL.md
1,445 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: ranked gap plan (contrast pair)
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,655 bytes
Read SKILL.md
# Coverage Gap Finder
Find the untested paths that would actually hurt, ranked by risk - never chase a line-coverage percentage. Coverage is necessary but not sufficient: a line can execute under test while nothing asserts its output, and a percentage says nothing about whether the uncovered 15% is dead config or the payment splitter. The costly mistake this skill prevents is a team burning a sprint raising 85% to 95% while the branch that corrupts money stays dark.
## Operating procedure
### Step 1: Gather inputs
- The coverage report - and whether it includes branch coverage. If it is line-only, regenerate before anything else (Step 2). No report means no prescription.
- Repo access for churn analysis, or at minimum the list of recently active files.
- The domain's crown jewels: where money moves, auth decisions happen, data is deleted or migrated. If the user cannot name them, propose a list and label it a guess.
- Release timeline, which sets how deep the ranked list must go.
### Step 2: Generate a real branch-coverage report
creator-content-calendar
When your agent uses it
Designs a sustainable publishing system for a solo creator - 3-4 content pillars, a cadence set at the creator's consistent floor, a fixed pillar rotation, a 30-day idea bank, and a monthly audit loop - delivered as a weekly calendar skeleton. Use when someone says "help me plan my content", "I keep falling off my posting schedule", "I'm launching a channel, what should my cadence be", or "audit my stalled content strategy". Do NOT use for a brand or social team's multi-platform posting grid with per-channel cadences and campaign dates - use social-content-calendar instead; do NOT use to fan one finished pillar piece into derivatives - use content-repurposing instead; do NOT use to write the individual posts or scripts - use youtube-script-writer or linkedin-post-writer instead.
What it loads into context
Frontmatter
180 tokensAlways available to the agent
SKILL.md
1,776 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Weekly calendar skeleton
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,564 bytes
Read SKILL.md
# Creator Content Calendar
A content calendar is a sustainability tool, not an idea list. Its job is to prevent decision fatigue, guarantee pillar balance, and make consistency achievable at the creator's real energy level - because the channel-killer is not bad content, it is the boom-bust cycle where an ambitious schedule collapses after three weeks and the creator disappears for two months. This skill produces a calendar structure the creator can hold through their worst week.
## Operating procedure
Run in order: cadence decisions made before pillars exist produce a schedule with nothing to fill it, and idea banks built before pillars produce ideas with no home.
### Step 1: gather inputs
1. Real weekly hours available for content, on a bad week - not an inspired one. This is the binding constraint.
2. Fixed constraints: job, family, travel rhythm, energy patterns (batch filmer vs. daily writer).
3. Primary channel and format (the calendar is single-creator; multi-channel derivatives come from repurposing, not extra slots).
4. Current output history: what they actually shipped in the last 8 weeks, not what they intended. Label any estimate a guess.crisis-comms-external
When your agent uses it
Drafts and sequences external communications during an incident or crisis - a first statement within the hour, acknowledge-what-you-know-without-speculating discipline, a single-spokesperson rule, channel sequencing from status page to email to social, and a legal-review escalation boundary, with severity-tiered statement templates. Use when a user says "we have an outage and customers are asking questions", "we had a data breach, what do we say publicly", "draft a public statement about this incident", "the story is blowing up on social, how do we respond", or "when do we need legal to review this statement". Do NOT use for routine technical status posts during normal-severity incidents - use status-page-update instead - for the internal writeup after resolution - use postmortem-writer instead - or for calming one angry customer - use refund-deescalation instead.
What it loads into context
Frontmatter
187 tokensAlways available to the agent
SKILL.md
2,561 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Concrete thresholds
- Statement templates by severity
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore12,079 bytes
Read SKILL.md
# Crisis Comms External
In a crisis, silence is a statement - and it says "we don't know" or "we're hiding something," both of which customers and journalists will fill in for you. The costly mistake this skill prevents is the double failure that defines botched crisis comms: waiting hours for complete information before saying anything, then over-speaking with speculation that later proves wrong and must be retracted. The discipline is the opposite pairing - speak fast, but say only what is verified.
## Operating procedure
Steps 1-3 happen in the first hour, in parallel where possible. The clock starts when the organization becomes aware customers are affected, not when the root cause is found.
### Step 1: Classify severity and open the log
Classify immediately, because severity drives everything downstream - template, channels, legal involvement:
- **SEV-A - Trust events:** data breach, security incident, safety issue, anything touching personal data or money. Legal review is mandatory before any external statement (see Step 6).
- **SEV-B - Broad service failure:** major outage or degradation affecting a large share of customers, visible enough to draw press or social attention.cross-platform-reformatter
When your agent uses it
Re-expresses one finished piece of content as the native-equivalent post on each target channel, holding the core idea constant while flexing length, structure, register, and conventions per platform. Use when you have a single written post, article, script, or message and need the same idea posted natively on LinkedIn, X, Instagram, TikTok, YouTube Shorts, or a newsletter. Do NOT use when fanning one long-form asset out into different asset types (clip, quote card, blog) - use content-repurposing instead; do NOT use when writing one fresh caption from a topic - use social-caption-writer instead.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,266 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Worked example
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,617 bytesSKILL.mdcore5,617 bytesSKILL.mdcore5,617 bytes
Read SKILL.md
# Cross-Platform Reformatter
Take one existing piece and ship the same idea as a native post on each requested channel - not the same text with hashtags swapped.
## Workflow
1. Extract the core: pull the single transferable idea plus its proof (the stat, story, or takeaway). This stays constant across every version; everything else is rebuilt around it.
2. Confirm the target channels. Reformat only the channels the user named; do not invent a full distribution stack. If they named none, ask which channels.
3. Rebuild - do not truncate - for each target, in its native grammar:
- LinkedIn: text-first, ~1300-2000 chars, short lines with white space, professional-but-human, one clear insight.
- X: a tight single post under 280, or a thread where each post stands alone; punchy, no corporate tone.
- Instagram: caption ~125-150 chars or a short narrative; the image/Reel carries the weight.
- TikTok: a spoken script with a verbal hook in the first two seconds; conversational, written to be said aloud.
- YouTube Shorts / Reels: the same script logic with on-screen text beats.
- Newsletter: longest form; keep the nuance the social cuts drop.
4. Apply native conventions per channel: hashtags 3-5 on Instagram, 1-2 on LinkedIn/X, a couple of discovery tags at most on TikTok; use features as intended (carousels/Reels on IG, document posts and polls on LinkedIn, threads and quote-posts on X); tune emoji density to the channel.csat-root-cause
When your agent uses it
Analyzes CSAT and NPS verbatims to surface root-cause themes and prioritize fixes by frequency and impact. Use when interpreting customer satisfaction data to inform product or process decisions.
What it loads into context
Frontmatter
60 tokensAlways available to the agent
SKILL.md
815 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Data preparation
- Coding scheme
- Prioritization matrix
- Presenting findings
- Thresholds that signal urgency
- Deliverable
- What to avoid
- Quality bar
Files
SKILL.mdcore3,796 bytes
Read SKILL.md
# CSAT Root Cause Analysis
Raw CSAT scores tell you that something is wrong. Verbatim comments tell you what and why. The goal of root-cause analysis is to turn hundreds of individual comments into a ranked list of fixable problems.
## Data preparation
Before analyzing, filter the dataset: remove blank verbatims, remove comments that mention only a specific agent by name (these are agent-performance signals, not product/process signals), and separate detractor verbatims (score 0-6 on NPS, 1-3 on CSAT) from promoter verbatims. Analyze detractors first - they drive churn.
## Coding scheme
Apply a two-level coding scheme. Level 1 is the broad category (product bug, slow response time, unclear communication, pricing confusion, onboarding friction, feature gap, policy frustration). Level 2 is the specific sub-theme within that category. Code each verbatim with exactly one Level 1 and one Level 2 code. If a verbatim contains multiple distinct complaints, split it into sub-comments before coding.
## Prioritization matrix
curriculum-mapper
When your agent uses it
Maps a course's scope and sequence across a term or year - units, weeks, standards coverage with introduced/developed/mastered notation, and Bloom's-level progression - and flags gaps and redundancy. Use when someone asks "map my curriculum for the year", "build a scope and sequence for this course", "audit my course for standards gaps", or is deciding unit order for a semester. Do NOT use for planning a single lesson - use lesson-plan-builder instead; for adapting one lesson to mixed readiness levels in one classroom, use differentiated-instruction.
What it loads into context
Frontmatter
128 tokensAlways available to the agent
SKILL.md
1,545 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: two map rows
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,357 bytes
Read SKILL.md
# Curriculum Mapper
Curriculum mapping makes the invisible visible: what is taught, when, in what order, and whether the sequence actually serves learning. The costly failure it prevents is discovered in April - a standard that was never taught, a unit taught twice by different teachers, or a year whose cognitive ceiling never rose above recall. A good map surfaces those problems in August, when they cost a planning hour instead of a cohort's outcomes.
## Operating procedure
Order matters: parameters constrain units, units constrain sequence, and standards alignment is only checkable once the sequence exists.
### Step 1: Gather parameters
Collect before mapping; label estimates as estimates:
- Grade level or course, and duration (semester, trimester, year).
- Instructional days actually available: start from the calendar total, subtract testing windows, assemblies, and predictable non-instructional days. Most planners overestimate by 10-15%; a 180-day calendar commonly yields 150-160 teachable days.customer-analytics
When your agent uses it
Turns transaction data into cohort retention, lifetime value, RFM segments, and churn predictions tied to concrete actions. Use when someone asks "what's our customer LTV", "build a cohort retention table", "which customers are about to churn", "how do repeat purchase rates look by cohort", or "who should we target with a win-back offer". Do NOT use for event-level product usage and feature adoption analysis - use product-analytics instead; for defining market segments for positioning and messaging - use segmentation-strategy instead; for designing the interventions that reduce churn once at-risk customers are identified - use churn-reduction instead; for step-by-step conversion drop-off - use funnel-analysis instead.
What it loads into context
Frontmatter
155 tokensAlways available to the agent
SKILL.md
1,769 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: retention triangle excerpt with reading
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,790 bytesSKILL.mdcore7,790 bytes
Read SKILL.md
# Customer Analytics
Transaction logs already contain the answers to "who stays, who spends, who is about to leave" - the failure mode is computing them wrong: cohort charts that include the current incomplete month, LTV built on revenue instead of margin, and churn models validated on shuffled data that collapse in production. This skill produces cohort, LTV, and churn outputs a finance or lifecycle team can act on directly.
## Operating procedure
Work the steps in order: cohorts establish the retention reality that LTV math depends on, and churn modeling is only worth doing after the churn definition survives the cohort view.
### Step 1: Gather inputs
Collect these first; apply defaults and label guesses as guesses.
1. Transaction table: customer_id, order_date, revenue, and cost or margin if available.
2. Business model: contractual (subscription) or non-contractual (e-commerce/retail). This decides the LTV method in Step 3.customer-success-qbr
When your agent uses it
Runs quarterly business reviews that prove delivered value, score account health, and open the expansion and renewal conversation - producing an agenda, a value-delivered slide format, a health-score definition, and an expansion script. Use when someone asks "how do I run a QBR", "prep my quarterly business review", "build a customer health score", "how do I bring up renewal in a QBR", or is preparing an executive business review for a key account. Do NOT use for designing the expansion motion and pricing itself - use expansion-revenue instead - or for rescuing an account already at risk of churning - use churn-reduction instead.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,929 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Cadence and segmentation
- Template: QBR agenda
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,352 bytes
Read SKILL.md
# Customer Success QBR
A QBR is not a product demo or a support recap. It is a business conversation that proves the value you delivered, aligns on the customer's goals, and opens the door to expansion and renewal. Run badly - usage stats to a champion-only room - it trains the customer that QBRs are skippable, and the renewal conversation starts cold at day 90. Run well, it is the single best retention tool you have.
## Operating procedure
Follow the steps in order: the room decides whether the meeting matters, the pre-work decides whether you have anything to say, and the value section must land before any expansion ask is credible.
### Step 1: Gather inputs
Collect these before building anything. Where a number is unverified, label it a guess and confirm it with the champion before the meeting.
1. The customer's stated goals from last quarter, in their exact words (from the prior QBR notes or kickoff doc). If none exist, the first agenda item becomes setting them.
2. Contract facts: ARR, renewal date, products owned, seats licensed vs. active.dark-mode-design
When your agent uses it
Build a dark theme with correct elevation, surface hierarchy, and contrast. Use when adding dark mode to a product, or when an existing dark theme looks flat, muddy or hard to read.
What it loads into context
Frontmatter
56 tokensAlways available to the agent
SKILL.md
728 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core mistakes to avoid
- Step 1: Establish surface elevation
- Step 2: Text opacity hierarchy
- Step 3: Color adaptation
- Step 4: Contrast targets
- Step 5: Elevation + color interaction
- Step 6: Tokenize for theming
- Step 7: Test in context
- Output
Files
SKILL.mdcore2,880 bytes
Read SKILL.md
# Dark Mode Design
Use this skill to design a dark theme that is more than an inverted light theme.
Dark mode has its own rules for depth, color, and contrast.
## Core mistakes to avoid
- **Pure black backgrounds** (`#000`). Use a dark gray (e.g. `#121212`) so shadows and
elevation are visible and OLED smearing is reduced.
- **Pure white text** (`#FFF`) on dark. Use ~87% opacity white for body text to cut
harsh contrast and eye strain.
- **Saturated colors** carried over from light mode. They vibrate against dark
backgrounds - desaturate and lighten brand colors for dark surfaces.
data-drift-monitor
When your agent uses it
Designs a production drift-monitoring plan for ML systems - statistical tests per feature type, PSI and KS alert thresholds, monitoring cadence, and evidence-based retraining triggers - including a runnable PSI calculator. Use when someone asks "how do I know if my model is drifting", "set up drift monitoring for this model", "should we retrain", or is investigating unexplained model performance degradation. Do NOT use for judging whether a model is good enough to ship in the first place - use model-evaluation-report instead; for tracking training runs use experiment-tracking; for documenting the model for consumers use model-card-writer.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,705 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: PSI calculator with output
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,404 bytesSKILL.mdcore7,404 bytes
Read SKILL.md
# Data Drift Monitor
Models trained on historical data degrade when the world changes. Drift monitoring is the early-warning system that distinguishes a model aging gracefully from one silently failing - and without thresholds tied to actions, it degenerates into dashboards nobody reads and alerts everybody mutes. This skill produces a monitoring plan where every alert has a named owner and a defined next step.
## Operating procedure
Work in order: drift types (Step 2) determine which tests apply (Step 3), and thresholds (Step 5) are meaningless until cadence and coverage (Step 4) define what is measured.
### Step 1: Gather inputs
1. The model's top features by importance (at minimum the top 10) and their types (continuous vs categorical).
2. Prediction volume and frequency - this sets monitoring cadence.
3. Ground-truth label lag: how long after a prediction the true label arrives. If unknown, label the estimate a guess.
4. The primary business metric the model serves and its current 30-day baseline.data-quality
When your agent uses it
Designs layered data quality checks across completeness, validity, consistency, uniqueness, timeliness, and accuracy, with severity tiers, freshness SLAs, and anomaly baselines wired into dbt and CI. Use when someone asks "how do I stop bad data reaching dashboards", "set up dbt tests for this model", "our pipeline loaded duplicate rows again", "what data quality checks should this table have", or "how fresh does this source need to be". Do NOT use for detecting distribution drift in ML features and predictions - use data-drift-monitor instead; for one-off exploration and profiling of a new dataset - use eda-playbook instead; for infrastructure and application telemetry - use observability-stack instead; for removing personal data from datasets - use pii-scrubber instead.
What it loads into context
Frontmatter
173 tokensAlways available to the agent
SKILL.md
1,747 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: check-spec template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,025 bytesSKILL.mdcore8,025 bytesSKILL.mdcore8,025 bytes
Read SKILL.md
# Data Quality Framework
Bad data costs most when it is discovered by the consumer - an executive quoting a dashboard built on a half-loaded table, or a model trained on duplicated rows. The costly mistake this skill prevents is the all-or-nothing test suite: either no checks, or hundreds of untiered checks that page someone nightly until they are all silenced. The output is a small, tiered check suite where every failure has a defined severity and action.
## Operating procedure
Severity tiering (Step 3) comes before implementation because a check without a defined response is future alert fatigue.
### Step 1: Gather inputs
1. The tables in scope, their primary keys, and who consumes them. Default scope: start with the 3-5 tables feeding the most-viewed dashboards or production models, not the whole warehouse.
2. Load cadence and expected arrival time per source.
3. Business rules that define a valid row (status enums, amount ranges, referential links).
4. Where checks will run: dbt is the default assumption below; the structure ports to any framework.data-room-builder
When your agent uses it
Use when a founder needs to prepare materials for investor due diligence. Triggers on "set up a data room", "what goes in a data room", "diligence checklist", "prepare for due diligence", "investors asked for our data room". Builds the organized, staged room that makes diligence fast and confidence-building.
What it loads into context
Frontmatter
74 tokensAlways available to the agent
SKILL.md
902 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- Structure the room in tiers
- Operating rules
- Anti-patterns
- Quality bar
- Deliverable
Files
SKILL.mdcore3,930 bytesSKILL.mdcore3,930 bytesSKILL.mdcore3,930 bytesSKILL.mdcore3,930 bytes
Read SKILL.md
# Data Room Builder
The data room is where momentum goes to die or to accelerate. A messy,
slow-to-assemble room signals an undisciplined company and gives a wavering
investor a reason to drift. A clean, staged room signals "this founder runs a
tight ship" and shortens the path to wired funds.
## When to use this skill
Build it during [[fundraise-readiness-audit]] (you should be able to produce it
in days - 3 to 5 working days is the standard for a seed-stage company with
clean books), and stage it for sharing once a term sheet is near. Confirmatory
diligence runs against it; see [[fundraise-closing-mechanics]].
data-story
When your agent uses it
Turns data and charts into a decision-driving narrative structured as headline finding, trend, implication, and recommended action - with finding-led chart titles, context for every number, annotation guidance, and honest flags on any conclusion the data cannot support. Use when someone says "turn these numbers into a story", "what's the takeaway from this data", "help me present these results to leadership", or has charts but no narrative. Do NOT use for compressing a long document into a one-pager - use executive-summary instead - or for running the analysis that produces the findings - use eda-playbook instead.
What it loads into context
Frontmatter
139 tokensAlways available to the agent
SKILL.md
1,053 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core principle
- Process
- The four-part structure
- Finding the story
- Presenting numbers
- Writing rules
- Anti-patterns
- Quality bar
- Output
Files
SKILL.mdcore4,893 bytes
Read SKILL.md
# Data Storyteller
You turn numbers and charts into a story that drives a decision. Data alone doesn't persuade; the narrative around it does. Your job is to find the "so what" and tell it.
## Core principle
Every dataset has a story, but it has to be found and framed. Lead with the insight, support with the data - not the other way around. An audience remembers the takeaway, not the spreadsheet.
## Process
1. Get the data/charts plus context: the audience, the decision at stake, and what they believe going in.
2. Find the one finding that matters most. Interrogate the data: what changed, what's surprising, what's the outlier, what's the trend.
3. Build the narrative: headline → evidence → implication → action.
data-table-design
When your agent uses it
Designs presentation tables that read correctly at a glance - numeric alignment, consistent precision, deliberate sort order, unit-labeled headers, and well-placed totals - and delivers a before/after redesign of the reader's table. Use when someone asks "make this table more readable", "how many decimals should I show", "should totals go at the top or bottom", "this report table is a mess", or is preparing tabular data for an exec deck or published report. Do NOT use for designing database tables, keys, and normalization - use database-schema instead; for narrative structure around the numbers - use data-story instead; for assembling the full multi-section document - use formatted-report-writer instead.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,447 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: before/after
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,708 bytes
Read SKILL.md
# Data Table Design
A table is not a spreadsheet dump. Every formatting choice either helps or hinders the reader's ability to compare, rank, and interpret - and a misaligned or falsely precise table gets numbers misread in the meetings that matter. The output of this skill is a table a reader can extract the right conclusion from in under ten seconds.
## Operating procedure
### Step 1: Gather inputs
1. The table as it stands (or the raw data), and where it will live: slide, printed report, dashboard, or scrolling web page - the medium changes the density rules in Step 5.
2. The reader's single most likely question. Every subsequent decision serves it. If the owner cannot name the question, that is the first problem to fix.
3. The action the reader takes from the table. Columns that inform no comparison or action are candidates for deletion.
4. Units and the true precision of the data - what the collection method actually supports, not what the export prints.
### Step 2: Set alignmentdatabase-schema
When your agent uses it
Designs normalized, constrained, migration-friendly relational schemas - entity modeling, key and type selection, indexes derived from real query patterns, and safe forward/rollback migrations. Use when someone asks "design a schema for X", "should I normalize or denormalize this", "what should my primary key be", "how do I add this column without downtime", or "why is my unique constraint broken with soft deletes". Do NOT use for tuning a slow query on an existing schema - use sql-query-optimizer instead; for picking indexes from a live workload, use index-advisor; for sizing database connection pools, use connection-pool-tuner; for range/hash partitioning decisions, use partition-planner.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,906 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast: the schema and the migration it costs you
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,635 bytes
Read SKILL.md
# Database Schema Designer
A schema is the one part of a system you cannot refactor with a find-and-replace: every shortcut taken at design time becomes a locked-table migration on a hot production database later. This skill produces schemas that stay correct and fast as data grows - invalid states unrepresentable, indexes matched to real queries, and every change shippable without downtime.
## Operating procedure
Follow the steps in order. Constraints come before indexes because constraints define correctness and indexes only define speed; migrations come last because they depend on both.
### Step 1: Gather inputs
Collect these before writing any DDL. If a number is a guess, label it a guess and proceed.
1. Entities and relationships - nouns of the domain and their cardinality (1:1, 1:N, M:N).
2. The top 5-10 queries by expected frequency, with their filter and sort columns. Indexes come from this list, not from intuition.dead-code-eliminator
When your agent uses it
Proves code is unreachable with converging evidence, then removes it and its tests, fixtures, and flags in reversible slices without breaking dynamic callers. Use when deleting suspected dead code, cleaning up after a migration or feature retirement, trimming a bloated module, or before estimating work on unfamiliar code. Do NOT use when you want to survey and rank an area's debt without committing to deletion - use find-tech-debt instead.
What it loads into context
Frontmatter
105 tokensAlways available to the agent
SKILL.md
894 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,084 bytes
Read SKILL.md
# Dead Code Eliminator
Delete code only after converging evidence proves it dead, then remove it in reversible slices that include its tests, fixtures, config, and flags.
## Workflow
1. **Scope one cluster.** Pick a single symbol, file, or feature to prove dead. Do not batch unrelated removals - a regression must point to one change.
2. **Gather static evidence.** Run the language's dead-code detector (ts-prune or knip for TS/JS, vulture for Python, `deadcode` for Go) and a dependency grapher (madge) or IDE "find usages". Record what each reports unreachable.
3. **Gather runtime evidence.** Check production coverage or profiling data over a representative window - at least 30 days for user-facing paths, 90+ days for code tied to monthly or quarterly jobs - and telemetry: was this endpoint, job, or branch hit this quarter? A symbol is a deletion candidate only when static analysis AND runtime evidence agree. One signal alone produces false positives that take down prod.
4. **Hunt the dynamic callers static tools miss.** Grep the symbol name as a *string* (reflection, string-keyed dispatch, serialization). Check DI containers, registries, ORM hooks, cron/queue workers, and feature-flag config. Search infra repos and other services. Public API surface and library exports are reachable by definition unless you control every consumer.
5. **Classify the kind of dead, because each is removed differently.** Unreachable (no path calls it) → delete. Unused (callable, never called) → delete. Redundant (duplicates a live path) → repoint callers first. Dormant-behind-a-flag (flag off for a cycle) → retire the flag first, then delete its branch.
6. **Soft-delete risky removals first.** Replace the suspect path with log-and-throw, or gate it behind a kill flag while keeping the code, for one full business cycle - 30 days minimum, 90 for anything touching monthly or quarterly processes. If nothing fires the log, proceed. For low-risk leaf code, skip to step 7.
7. **Delete in a focused PR.** Remove the code AND its tests, fixtures, config, flags, and now-orphaned imports. Keep the diff under roughly 400 changed lines so it is reviewable in one sitting; a bigger cluster gets split into multiple PRs. Let CI and the type-checker catch references the search missed. No formatting or refactors in the same diff.
## Deliverabledebt-payoff-planner
When your agent uses it
Builds a sequenced multi-debt payoff plan - avalanche or snowball chosen by explicit decision rules - with the rollover schedule, total-interest math, and a debt-free date. Use when someone asks "which debt do I pay first", "avalanche or snowball", "how do I get out of credit card debt", or "when will I be debt-free". Do NOT use for sizing the cash cushion that prevents new debt - use emergency-fund-planner instead; for building the monthly budget that produces the extra payment - use budget-builder; for the overall save-vs-invest-vs-payoff ordering - use financial-planner.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,627 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: avalanche vs snowball on the same debts
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,155 bytes
Read SKILL.md
# Debt Payoff Planner
Carrying multiple debts without a sequenced plan means paying more interest than necessary and losing momentum. This skill produces a concrete, ordered repayment plan with a debt-free date. The costly mistake it prevents is spreading extra payments evenly across all debts - which feels fair and maximizes total interest paid.
## Operating procedure
### Step 1: Inventory every debt
Collect four fields per debt: creditor name, current balance, APR, minimum monthly payment. Include everything - credit cards, personal loans, medical debt, student loans, car loans. Exclude the mortgage from the active payoff list; it is managed separately. Label estimated APRs as guesses and confirm from statements.
Also collect gross monthly income and compute **debt-to-income** (total monthly debt payments ÷ gross monthly income). Under 36% is workable. Above 43% is the red line: the plan likely needs professional restructuring, not just sequencing - see Escalation. If unsecured debt exceeds annual income, escalate immediately.
### Step 2: Find the extra payment
decision-making
When your agent uses it
Runs a structured decision process for high-stakes choices - reversibility classification, weighted decision matrices, expected value with explicit probabilities, pre-mortems, second-order thinking, and 10/10/10 - and produces a documented recommendation plus a decision-journal entry. Use when someone asks "should I take this job or stay", "how do I decide between these two paths", "I keep going back and forth on a big decision", or "help me think through whether to relocate". Do NOT use for evaluating a specific large purchase like a house, car, or equipment - use big-purchase-decision instead.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,665 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,267 bytesSKILL.mdcore7,267 bytes
Read SKILL.md
# Decision Making
Most bad decisions are not caused by bad reasoning - they are caused by deciding the wrong question, rationalizing a pre-chosen favorite, or spending a week of analysis on a choice that could be undone in an afternoon. This process matches the depth of analysis to the stakes, forces criteria to be written before options are scored, and leaves a journal entry so judgment can be calibrated later.
## Operating procedure
Order matters: classification comes first because it decides whether the rest of the process runs at all, and criteria must be locked before any option is scored - scoring first invites rationalizing a favorite.
### Step 1: Gather inputs
Collect before analyzing. Label any estimate the user cannot source as a guess.
- The decision, stated in one sentence with a deadline. Default deadline if none exists: set one now - endless analysis is itself a costly choice.
- The real options on the table. Push for at least three; a binary framing usually hides a third option (negotiate, delay, pilot, partial commitment).deep-research
When your agent uses it
Runs a thorough multi-source research process - decompose the question, gather independent sources, cross-check, synthesize - and delivers a structured, cited brief with confidence levels and gaps. Use when someone asks "research this topic in depth", "give me everything on X", "what does the evidence say about Y", or needs a broad question answered from secondary sources with citations. Do NOT use for designing a study that collects new data - use primary-research instead; for distilling an already-gathered pile of 20+ sources into themes, use research-synthesis; for verifying one discrete claim, use fact-checker.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,195 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked example (findings-table format)
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,131 bytesSKILL.mdcore6,131 bytesSKILL.mdcore6,131 bytesSKILL.mdcore6,131 bytesSKILL.mdcore6,131 bytes
Read SKILL.md
# Deep Research
Go wide, then deep: triangulate across independent sources and report what was actually found, including the gaps. The costly failure this skill prevents is the confident brief built on an echo - a dozen articles that all trace to one blog post or one press release, presented as consensus. "Not found" is a legitimate finding; a smooth answer papering over thin evidence is not.
## Inputs to collect
1. The research question, restated until the user confirms it. Ambiguity here multiplies through every later step.
2. What decision or output the research feeds. A brief informing a purchase differs from one informing a publication.
3. Scope boundaries: geography, industry, population, and what is explicitly out of scope.
4. Depth and deadline. Default: 3-6 sub-questions, at least two independent sources per sub-question.
5. Known sources or prior work to build on. Treat supplied sources as leads to appraise, not settled evidence.
Label any assumed scope as an assumption.
deep-work-planner
When your agent uses it
Designs a daily schedule that protects deep work - classifies tasks as deep or shallow, places 2-4 focus blocks of 60-120 minutes at peak energy hours, batches shallow work into fixed windows, and defines each session with a goal, definition of done, distraction plan, and first action, plus startup/shutdown rituals and a weekly scorecard. Use when someone says "I can't get any focused work done", "plan my day around deep work", "how do I stop context switching", or wants to defend calendar time for demanding work. Do NOT use for processing an email backlog - use inbox-zero instead - or for a full task-capture system - use gtd-system instead.
What it loads into context
Frontmatter
159 tokensAlways available to the agent
SKILL.md
739 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Step 1: Classify your tasks
- Step 2: Schedule deep blocks
- Step 3: Batch the shallow
- Step 4: Design the session
- Focus rituals
- Energy management
- Weekly scorecard
- Anti-patterns
Files
SKILL.mdcore3,171 bytes
Read SKILL.md
# Deep Work Planner
Deep work is cognitively demanding work done without distraction. It produces
disproportionate value. This skill helps you protect and design it.
## Step 1: Classify your tasks
Sort the day's tasks into two lists:
- **Deep** - requires sustained focus, creates new value, hard to replicate
(writing, designing, coding hard logic, strategy).
- **Shallow** - logistical, low-concentration, often interrupt-driven
(email, Slack, scheduling, quick reviews).
demo-script
When your agent uses it
Builds a product demo script around the buyer's discovered pain - a 60-second situation recap opening, three to four before/after "moments" mapped to the buyer's metric, and a close that ends before the slot does. Use when someone asks "help me script this demo", "how should I structure tomorrow's product walkthrough", "turn these discovery notes into a demo plan", or before any live or recorded demo. Do NOT use for staging and directing the demo environment, data, and rehearsal - use product-demo-director instead. Do NOT use when discovery hasn't happened yet - run discovery-call-prep first, because a demo without known pain is a feature tour.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,647 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Beat structure (copy-paste skeleton)
- Deliverable
- Do NOT
- Quality bar
- Escape hatch
Files
SKILL.mdcore7,269 bytes
Read SKILL.md
# Demo Script
A feature tour is not a demo. Buyers do not care about features - they care about outcomes, and the job of a demo is to make the buyer see themselves solving a specific problem they described in discovery. The costly failure this skill prevents is the tour that fills the whole slot, impresses the presenter, and leaves the buyer with no reason to take a next step.
## Operating procedure
The order is fixed: inputs gate everything (no discovery, no demo), moments are selected before the open is written (the recap promises what the moments deliver), and pacing is set last.
### Step 1: Gather inputs - the three-fact gate
Pull the discovery notes and extract:
1. The primary pain the buyer named, in their words.
2. The current workaround they use today.dependency-risk-audit
When your agent uses it
Audits third-party dependencies for exploitable CVEs, abandonment, license exposure, and supply-chain hygiene, and delivers a ranked findings report with a remediation order. Use when someone asks "is this package safe to add", "audit our dependencies", "npm audit is screaming, what actually matters", "can we use this GPL library", or is preparing a security review or vendor questionnaire. Do NOT use for triaging vulnerabilities in first-party code or a full CVE queue - use vulnerability-triage instead; for reviewing the code you wrote for security flaws - use secure-code-review instead; for how secrets are stored and rotated - use secrets-hygiene instead; for assembling compliance evidence - use soc2-evidence-helper instead.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,733 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast: a finding writeup
- Deliverable
- Do NOT
- Escalation
- Quality bar
Files
SKILL.mdcore7,705 bytesSKILL.mdcore7,705 bytes
Read SKILL.md
# Dependency Risk Audit
Third-party packages are the most common source of supply-chain compromise and known-vulnerability exploitation; treat dependency selection as a security decision, not a convenience one. The costly failure this skill prevents is on both ends: shipping a reachable critical CVE because the audit output was noise nobody read, and burning a sprint patching findings in code paths that can never execute.
## Operating procedure
### Step 1: gather inputs
Collect (label guesses as guesses):
1. Ecosystem and manifest/lockfile (package.json + lockfile, requirements/poetry, Gemfile.lock, go.mod).
2. Deployment context: server-side, client-side, CLI-only, or build-time only - this decides reachability.
3. License posture: proprietary/commercial code, or open source (decides copyleft tolerance).
4. Whether this is a point decision (one new package) or a full audit (whole tree).design-critique
When your agent uses it
Runs a structured design critique of a screen, flow, or component across three dimensions - jobs-to-be-done, clarity, and delight - and delivers a severity-ranked critique table with one prioritized recommendation. Use when someone asks "critique this design", "review this mockup before we build it", "this flow feels off but I can't say why", "which design debt matters", or a new design or redesign is up for review. Do NOT use for verifying a built implementation against its spec - use design-qa-checklist instead; for planning moderated usability sessions, use usability-test-plan.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,445 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use
- Inputs to collect
- The three dimensions
- Output format
- Rules of engagement
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,637 bytes
Read SKILL.md
# Design Critique
A vague reaction ("I don't like it") wastes a design review; the team leaves with hurt feelings and no direction, then ships the same problems. A rigorous critique anchors every note to a dimension, a severity, and a concrete fix - so a designer or PM walks out knowing exactly what to change and in what order. This skill runs pre-build, on mockups and prototypes, where a fix costs minutes instead of sprints.
## When to use
- A new design or redesign is up for review.
- A flow feels "off" but no one can articulate why.
- You need to prioritize design debt against shipping goals.
## Inputs to collect
Gather three things first. If any is missing, ask for it before critiquing - a critique without them degenerates into taste.
design-handoff-doc
When your agent uses it
Writes a complete design handoff document covering components, design tokens, interaction states, and edge cases for engineering implementation. Use when handing off a feature or screen to developers.
What it loads into context
Frontmatter
54 tokensAlways available to the agent
SKILL.md
1,167 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Document Structure
- Component Inventory
- Design Tokens
- Interaction States
- Edge Cases and Constraints
- Acceptance Criteria Alignment
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore5,575 bytesSKILL.mdcore5,575 bytes
Read SKILL.md
# Design Handoff Doc
A handoff document is the single source of truth for what is being built. It replaces Slack threads, live walkthroughs, and guesswork. Write it for an engineer who joins the project the day implementation starts and has no prior context.
## Document Structure
Every handoff doc contains these sections in order:
1. Overview: what the feature does and why it exists (2-4 sentences max).
2. Design file link: direct link to the specific page or frame, not the project root.
3. Component inventory: list of every new or modified component.
4. Design tokens used: all color, spacing, typography, and radius tokens referenced.
5. Interaction states: every state for every interactive element.
6. Edge cases and constraints: what can go wrong and what the design does about it.
7. Out of scope: explicit list of what is NOT being built in this iteration.design-qa-checklist
When your agent uses it
Runs an ordered design QA pass over an implemented UI - layout, type, color, states, motion, content - against the design spec, with concrete tolerances and a filed defect list. Use when someone asks "QA this build against the designs", "does the implementation match Figma", "is this ready for design sign-off", or before releasing a UI change. Do NOT use for critiquing the design itself - use design-critique instead; for a full WCAG conformance audit, use accessibility-audit; for writing the spec the build is checked against, use design-handoff-doc.
What it loads into context
Frontmatter
142 tokensAlways available to the agent
SKILL.md
1,563 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked defect entries
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,938 bytesSKILL.mdcore6,938 bytes
Read SKILL.md
# Design QA Checklist
Design QA catches drift between design intent and shipped code before it compounds into a product that looks 90% right everywhere and fully right nowhere. The costly failure is an unordered spot-check: motion gets polished while the grid underneath is wrong, so half the findings are invalidated by the layout fix. Run the pass in order - layout first, because layout errors move everything downstream - and file every deviation with expected vs. actual values, not adjectives.
## Operating procedure
### Step 1: Gather inputs
- The design source of truth (Figma file, spec, or design-handoff-doc output) and the build URL or branch.
- The token set (spacing scale, type scale, color tokens, motion tokens). If none exists, note that and QA against explicit spec values; flag "no token system" as its own finding.
- Supported breakpoints and the narrowest supported viewport (default: 320px).
- Whether dark mode and motion specs exist (default: assume light mode only, no motion spec - label as assumption).
### Step 2: Layout passdesign-token-system
When your agent uses it
Build a three-tier token architecture (global, semantic, component) with theming. Use when setting up design tokens, adding themes or dark mode to a design system, or untangling hard-coded colors and spacing.
What it loads into context
Frontmatter
62 tokensAlways available to the agent
SKILL.md
739 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The three tiers
- Why three tiers
- Theming with token sets
- Naming convention
- Token categories to cover
- Tooling
- Guardrails
- Output
Files
SKILL.mdcore3,255 bytesSKILL.mdcore3,255 bytes
Read SKILL.md
# Design Token System
Use this skill to design a scalable token architecture that survives rebrands,
theming, and growth - instead of hard-coded values sprinkled through the codebase.
## The three tiers
### Tier 1: Global (primitive) tokens
Raw values with no meaning attached. The full palette of options.
```js
const global = {
blue500: '#3B82F6',differentiated-instruction
When your agent uses it
Adapts an existing lesson or unit for mixed readiness levels using tiered assignments, flexible grouping, and scaffolds that fade - without lowering the learning objective. Use when someone says "differentiate this lesson", "I have students at three levels in one class", "add scaffolds and extensions to this unit", or needs one lesson to reach struggling and advanced learners at once. Do NOT use for writing a lesson from scratch - use lesson-plan-builder instead; for sequencing units across a term or auditing standards coverage, use curriculum-mapper.
What it loads into context
Frontmatter
127 tokensAlways available to the agent
SKILL.md
1,453 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: one task, three tiers
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,465 bytes
Read SKILL.md
# Differentiated Instruction
Differentiation is not creating separate lessons for every student. It is designing one lesson with intentional entry points, flexible pathways, and multiple ways to demonstrate mastery - same objective for all, different paths to it. The costly failure it prevents is quiet tracking: struggling students fed only low-complexity work are being denied grade-level learning while the paperwork says they were "supported."
## Operating procedure
### Step 1: Gather inputs
Collect before modifying anything; label unknowns as guesses to verify with a formative check:
- The original lesson or unit, its learning objective, and the core task as written.
- The readiness spread, from recent formative assessment data - not historical placement or last year's labels. If no data newer than three weeks exists, run a short formative probe first; differentiating on stale data re-creates the ability groups it is supposed to replace.
- Known learning needs (IEP/504 accommodations, language support) and rough group sizes per readiness band.
- What the teacher can actually manage: number of simultaneous versions, para/aide availability, time to prepare.discovery-call-prep
When your agent uses it
Prepares a rep for a B2B discovery call - targeted company research, a falsifiable hypothesis stack, a layered question plan with MEDDICC and SPICED coverage, and explicit call goals, delivered as a filled prep sheet. Use when someone asks "help me prep for a discovery call", "what should I ask this prospect tomorrow", "build a question plan for my first call with this account", or before any first or second call with a prospect. Do NOT use for scripting the pitch and close of an already-qualified deal - use closer-sales-script instead. Do NOT use for building the outbound sequence that books the call - use outreach-sequence-designer instead.
What it loads into context
Frontmatter
155 tokensAlways available to the agent
SKILL.md
1,815 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Talk-time and in-call discipline
- Template: discovery prep sheet
- Deliverable
- Do NOT
- Quality bar
- Escape hatches
Files
SKILL.mdcore7,960 bytes
Read SKILL.md
# Discovery Call Prep
A discovery call only earns the next step if the rep walks in with sharp hypotheses and leaves with the information needed to qualify and tailor the pitch. The costly failure this skill prevents is the "friendly call that goes nowhere": forty-five pleasant minutes, no pain quantified, no decision process mapped, no next step on the calendar - a slot burned and a deal that silently dies.
## Operating procedure
Follow the steps in order. Research must precede hypotheses (hypotheses come from signals, not imagination), and hypotheses must precede the question plan (each question exists to confirm or kill a hypothesis).
### Step 1: Gather inputs
Collect from the rep before doing anything else. Where an input is missing, use the default and label it a guess.
1. Prospect company name and the contact(s) on the call, with titles.
2. Call number (first or second) and how the meeting was sourced (inbound, outbound, referral) - inbound calls start with more permission to go deep.docker-compose-wizard
When your agent uses it
Produces production-minded docker-compose configurations - pinned images, healthchecks with real intervals, health-gated startup ordering, named volumes, explicit networks, and env-file secrets - plus the matching .env.example. Use when someone asks "write a compose file for this stack", "why does my app start before the database is ready", "how do I add a healthcheck", "my compose setup works on my machine but not in CI", or is reviewing a docker-compose.yml before it ships. Do NOT use for orchestrating across multiple hosts or cluster workloads - use kubernetes-basics instead; for provisioning the underlying cloud infrastructure - use terraform-expert instead; for CI pipelines that build and push the images - use github-actions instead; for auditing how secrets are stored and rotated - use secrets-hygiene instead.
What it loads into context
Frontmatter
183 tokensAlways available to the agent
SKILL.md
1,671 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,133 bytes
Read SKILL.md
# Docker Compose Wizard
Write compose files that are reproducible and safe to run, not just ones that start. The costly mistake this skill prevents is the works-on-my-machine file: `latest` tags that drift, an app that races its database on startup and crash-loops in CI, and state in anonymous volumes that vanishes on `down`.
## Operating procedure
### Step 1: gather inputs
Collect before writing (label guesses as guesses):
1. The services and their images or build contexts.
2. Which services hold state (databases, queues, caches) - these get named volumes.
3. Startup dependencies: who must be healthy before whom.
4. Which ports must reach the host; everything else stays internal.document-template-system
When your agent uses it
Builds a maintained library of reusable document templates for a team - frequency-based selection, four-part template anatomy, style and naming conventions, and a quarterly maintenance protocol. Use when someone asks "set up document templates for my team", "our docs are all formatted differently", "create a standard template for status reports or proposals", or "how do we keep templates from going stale". Do NOT use for designing Notion databases and their properties - use notion-database instead. For writing one specific process document rather than a template system, use process-doc; for visual brand rules, use brand-guidelines.
What it loads into context
Frontmatter
137 tokensAlways available to the agent
SKILL.md
1,674 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: a template built to the anatomy
- Deliverable
- Do NOT
- Quality bar
- Related skills
Files
SKILL.mdcore7,728 bytesSKILL.mdcore7,728 bytes
Read SKILL.md
# Document Template System
A template is not a filled-out document with blanks - it is a decision made once so the team never makes it again. The costly failure is the over-built system: twenty templates nobody uses, each slightly stale, until authors quietly revert to copying last quarter's doc and the formatting drift returns. The best template systems are small, self-explanatory, and maintained by one named person.
## Operating procedure
Audit before building, build before standardizing, standardize before naming - each step's output is the next step's input.
### Step 1: Audit what the team actually produces
List the top 10 document types by frequency (count real docs from the last quarter; do not poll people on what they think they write). Build templates only for the top 5. Attempting to template everything produces a system nobody uses - every template below the frequency line is pure maintenance debt.
For each of the top 5 candidates, answer three questions before designing anything:
donor-communications
When your agent uses it
Builds a small nonprofit's individual-donor communication system - segmenting donors into major, recurring, first-time, and lapsed with different treatment for each, thanking within 48 hours of every gift, structuring impact stories around one beneficiary with the donor as hero, writing appeal letters, and holding a 3:1 impact-to-ask calendar ratio. Use when a nonprofit leader says "our donors are lapsing", "write our year-end appeal", "how often should we email donors", "write a thank-you letter for a major gift", or "we only contact donors when we ask for money". Do NOT use for winning back lapsed donors with a full re-engagement sequence - use win-back-campaign for the sequencing mechanics - and do NOT use for foundation funder communications - use grant-loi-writer instead.
What it loads into context
Frontmatter
175 tokensAlways available to the agent
SKILL.md
2,101 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: segment treatment grid
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore9,818 bytes
Read SKILL.md
# Donor Communications
Most small nonprofits contact individual donors only when they want money, and then wonder why second gifts never come - when the sector-wide reality is that fewer than half of first-time donors ever give again, and the cheapest revenue any nonprofit has is a retained donor. This skill builds the communication system that keeps donors giving: segmentation, fast gratitude, impact stories, disciplined appeals, and a calendar that earns the right to ask.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students. Individual donors provide roughly $120,000 of the budget alongside the grant pipeline (see grant-prospect-researcher). Its outcome numbers are the canonical ones from the pack's logic model: 72 percent of students gained a grade level in reading last year - the same figure impact-report-builder publishes.
## Operating procedure
### Step 1: Segment the donor file
Four segments, four treatments. Sort every donor into exactly one:
- Major: the top of the file - for an organization Bright Futures' size, gifts of $1,000+ or the top 10 percent of donors. Treatment: personal. Handwritten or personally signed thanks, a phone call from the ED for gifts at this level, one-to-one updates, invited to see the program in person.
- Recurring: monthly or committed annual givers. Treatment: protect the relationship - never resolicit them in general appeals as if their gift were invisible; send an annual "your $25/month did this" cumulative-impact note.due-diligence-checklist
When your agent uses it
Runs structured diligence on an investment or acquisition target across four workstreams - commercial, financial, technical, and legal - and produces a red-flag register split into deal-killers versus manageable risks with a go/no-go recommendation. Use when someone asks "what should I check before investing", "run diligence on this company", "is this deal clean", or is preparing to wire money into a startup or acquisition. Do NOT use for decoding the clauses of a term sheet - use term-sheet-explainer instead; for negotiating terms, use term-sheet-negotiation; for building the target's data room from the company side, use data-room-builder.
What it loads into context
Frontmatter
153 tokensAlways available to the agent
SKILL.md
1,569 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Red-flag register template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,410 bytes
Read SKILL.md
# Due Diligence Checklist
Diligence exists to disconfirm the thesis, not confirm it: find the reason not to invest before the money moves. The costly mistake it prevents is confirmation bias - founders are persuasive, decks are polished, and the investor who only looks for supporting evidence reliably finds it. Enter assuming the deal is flawed and go hunting for the flaw; diligence is where skepticism is earned.
## Operating procedure
Run the four workstreams in this order. Commercial comes first because a broken market thesis makes the other three moot; legal comes last because its landmines are cheapest to check once the deal still looks alive.
### Step 1: Gather inputs
Collect before starting, and label every unverified founder claim as a claim:
1. The written investment thesis - one paragraph stating why this deal wins. Every finding gets tied back to it.
2. The data room or document set: financials, bank statements, cap table, material contracts, customer list.e2e-scenario-author
When your agent uses it
Converts an acceptance criterion or user story into one maintainable Playwright or Cypress end-to-end test that reads like the journey it covers. Use when turning a Given/When/Then acceptance criterion or user story into a browser test, when asked to "write an E2E test for this AC", or when a feature's critical journey (login, checkout, signup) needs a single automated scenario. Do NOT use when writing general Playwright tests, configuring the runner, or fixing existing test flakiness - use playwright-testing or flaky-test-detangler instead.
What it loads into context
Frontmatter
135 tokensAlways available to the agent
SKILL.md
939 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,315 bytes
Read SKILL.md
# E2E Scenario Author
Turn one acceptance criterion into one browser test that reads like the user story it came from and survives every refactor that does not change user-visible behavior.
## Workflow
1. **Pin the journey.** Take the single acceptance criterion and map its Given/When/Then to setup, action, assertion. One scenario covers one journey (e.g. "a returning user checks out a cart with a coupon"). If the AC implies several behaviors, write several tests. Name each test after the behavior, not the page.
2. **Seed state via the back door.** In setup, create the test's own data through an API request or a test-only route - never drive the UI through registration or other flows just to reach the start state. Back-door seeding typically cuts setup from 30+ seconds of UI clicking to under 2 seconds. Reset auth and storage so the test owns a fresh, isolated account.
3. **Drive the action through a page object.** Wrap each page or component in a page object exposing intent-level methods (`loginPage.signIn(user)`) so selectors live in one place and the test body speaks the domain. In Playwright, inject the authenticated page and shared setup via fixtures; in Cypress, use custom commands. The test body contains no raw selectors.
4. **Select by role and accessible name.** Inside the page object, use `getByRole('button', { name: 'Place order' })`, `getByLabel`, `getByText` - selectors that mirror how a user finds elements and double as accessibility checks. When a stable hook is genuinely unavoidable, add an explicit `data-testid`.
5. **Assert on user-visible state, waiting on signals.** Assert the outcome the user would see. Rely on auto-waiting: Playwright web-first `expect` assertions and Cypress commands retry until the condition holds. Wait for a real signal - a network response, a URL change, a visible element - never a fixed delay. Treat the runner's default timeout (Playwright's 30s) as a ceiling for pathological slowness, not a target the test should ever approach.
6. **Confirm isolation.** Verify the test shares no account or data with any other scenario and can run in parallel.
## Deliverable
economic-analysis
When your agent uses it
Applies microeconomic frameworks - supply and demand, elasticity, marginal analysis, market structure, externalities - to a business decision and produces a recommendation with explicit assumptions and a sensitivity analysis showing which assumption flips the answer. Use when someone asks "what happens to revenue if we raise prices", "will profits in this market hold up", "how will competitors respond to this move", or needs the economic logic behind a pricing, entry, capacity, or regulation question. Do NOT use for setting a specific price point or packaging tiers - use pricing-strategy or saas-pricing instead; for putting a dollar figure on market size, use market-sizing instead.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,576 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: sensitivity table
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,639 bytes
Read SKILL.md
# Economic Analysis
Economic reasoning makes the incentives and trade-offs behind a decision explicit - and shows which assumption the whole conclusion hangs on. The costly failure this skill prevents is the confident recommendation built on an unstated elasticity guess: small changes in that one number flip "raise prices" into "lose the market," and nobody notices until revenue does.
## Inputs to collect
1. **The decision**: the specific choice being made (raise price, enter market, add capacity, respond to a rival).
2. **The actors**: buyers, sellers, regulators - and who the analysis is for.
3. **The good or service** and its closest substitutes and complements.
4. **Market structure evidence**: number of competitors, differentiation, entry barriers.
5. **Any real data**: past price changes and volume responses, cost breakdowns, competitor prices. Real data beats framework defaults; where data is absent, state the assumed value and label it a guess.
## Operating procedure
eda-playbook
When your agent uses it
Runs a structured exploratory data analysis on a new or suspect dataset - schema audit, target analysis, feature profiling, missingness patterns, and leakage checks - ending in a written decision log. Use when someone says "I just got this dataset, where do I start", "my model metrics look too good", "audit this data before we model it", or is debugging unexpected model behavior. Do NOT use for writing the transformation code itself - use pandas-expert instead; for ongoing production data monitoring use data-quality; for constructing model features after EDA use ml-feature-engineering; for answering a one-off business question from a database use sql-to-insights.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,856 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: decision log template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,299 bytesSKILL.mdcore8,299 bytesSKILL.mdcore8,299 bytesSKILL.mdcore8,299 bytes
Read SKILL.md
# EDA Playbook
Skipping EDA before modeling is the single most common source of silent failures in ML pipelines: leakage that makes offline metrics fictional, structural missingness treated as random, a target definition nobody actually agreed on. This playbook is a fixed-order audit that surfaces those problems while they cost hours instead of quarters - and its output is a decision log, not a pile of plots.
## Operating procedure
Run the sections in order: schema problems invalidate the target analysis, and the target definition must be pinned before any feature-target relationship means anything.
### Step 1: gather inputs
- The dataset, its supposed grain (one row = ?), and the expected row count. If the provider cannot state either, derive them and label the result a guess.
- The target column and its **written definition** - precisely what event, over what window, measured when. Ambiguity here cascades everywhere; do not proceed on a verbal shrug.
- What information will exist **at prediction time**. This single fact powers every leakage check later.
- Any known collection changes (new form fields, pipeline migrations) that could create structural breaks.email-deliverability
When your agent uses it
Audits and protects sender reputation and inbox placement for permissioned marketing and lifecycle email - SPF, DKIM, and DMARC authentication, domain warmup, bounce and complaint thresholds, and engagement sunsetting - and delivers a scored deliverability audit checklist with a remediation order. Use when someone asks "why are my newsletters going to spam", "how do I set up SPF, DKIM, and DMARC", "is my bounce rate too high", "should I send marketing from a subdomain", or is launching a new sending domain or auditing an existing opted-in email program. Do NOT use for cold outbound infrastructure - lookalike sending domains, mailbox pools, per-mailbox volume caps - use cold-email-deliverability instead; do NOT use to write the emails themselves - use email-drip-builder or email-newsletter-pro instead.
What it loads into context
Frontmatter
187 tokensAlways available to the agent
SKILL.md
1,918 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: deliverability audit checklist
- Deliverable
- Do NOT
- Quality bar
- Scope and escalation
Files
SKILL.mdcore8,860 bytesSKILL.mdcore8,860 bytesSKILL.mdcore8,860 bytesSKILL.mdcore8,860 bytes
Read SKILL.md
# Email Deliverability
Sender reputation takes months to build and days to destroy, and most teams discover a deliverability problem only after open rates have already cratered. This skill audits and protects inbox placement for permissioned marketing and lifecycle email - authentication, warmup, list hygiene, and engagement - and returns a scored audit with fixes in dependency order. It exists to catch the silent killers (an unenforced DMARC policy, a stale segment, a skipped warmup) before an inbox provider does.
## Operating procedure
Work the steps in order: authentication failures poison every downstream metric, so fix identity before hygiene, and hygiene before engagement.
### Step 1: Gather inputs
Collect these before auditing anything. If a number is estimated, label it a guess and verify against the ESP dashboard before the audit ships.
- Sending domain(s), and whether marketing and transactional mail share one identity.
- ESP(s) in use; shared or dedicated sending IP.email-drip-builder
When your agent uses it
Designs automated onboarding and activation drip sequences with behavior-gated sends, exit triggers on conversion, and per-email jobs and timing. Use when someone asks "build an onboarding email sequence", "design a drip campaign for new signups", "why isn't our onboarding flow activating users", or is extending an existing drip series. Do NOT use for recurring content newsletters - use email-newsletter-pro; for recovering abandoned checkouts use abandoned-cart-sequence; for re-engaging dormant users use win-back-campaign; for cold outreach to prospects use outreach-sequence-designer or cold-email-craft.
What it loads into context
Frontmatter
137 tokensAlways available to the agent
SKILL.md
1,484 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact - 5-email activation drip (project-management SaaS; activation = first project shared with a teammate)
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,834 bytes
Read SKILL.md
# Email Drip Builder
A drip sequence is a contract with the user: they gave their attention, and the sequence returns it with value at the right moment. Sequences that read like mail merges get unsubscribes; sequences timed to behavior build habits. The costly mistake this skill prevents is the fixed-timer drip that keeps selling activation to users who already activated - the fastest way to teach a new user that these emails can be ignored.
## Inputs to collect
1. **The activation milestone** - the one action that means the user is activated (first invoice sent, first project shared). This is the exit trigger for the entire sequence. If undefined, define it first - a drip without an exit condition is a leak, not a funnel. Pull it from lifecycle-journey-map if that map exists.
2. **The intermediate milestones** - 2-4 steps between signup and activation, each trackable as an event.
3. **Current activation rate and where users stall** - determines which email gets the most work.
4. **Independently useful assets** - tips, benchmarks, shortcuts, case studies each email can deliver.
5. **Sending constraints** - ESP capabilities for event-based triggers vs. timers; if only timers exist, flag the sequence as degraded and say which rules cannot be enforced.
## Operating procedure
email-newsletter-pro
When your agent uses it
Write recurring email newsletters - subject lines that earn the open, preview text that extends them, a body structure readers finish, and cadence rules that keep the list healthy. Use when someone asks "write my newsletter", "give me subject line options", "why is my open rate dropping", "should I send weekly or monthly", or "help me structure this issue". Do NOT use for automated onboarding or behavior-triggered sequences - use email-drip-builder instead. Do NOT use for inbox placement and spam-folder problems - use email-deliverability instead.
What it loads into context
Frontmatter
129 tokensAlways available to the agent
SKILL.md
1,725 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Newsletter skeleton template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,554 bytes
Read SKILL.md
# Email Newsletter Pro
A newsletter competes with the whole inbox. The costly mistake is optimizing the body while the subject line loses the open, or winning opens with clickbait the body cannot pay off - both burn the asset that matters most: the reader's habit of opening. Earn the open, reward it, and protect the cadence that builds the habit.
This skill covers recurring broadcast issues sent to a list. Automated sequences triggered by signup or behavior belong to email-drip-builder; if mail is landing in spam, fix that first with email-deliverability.
## Operating procedure
### Step 1: Gather inputs
Collect before writing. Where the user does not know a number, use the default and label it a guess.
1. Audience and what they signed up for (the promise made at capture).
2. This issue's single core idea - one sentence. If they offer three, make them pick one; the others become future issues or secondary links.email-triage
When your agent uses it
Triages a batch of emails into delete, do-now, delegate, or defer decisions using the 2-minute rule, returning drafted replies, delegation forwards, and dated tasks. Use when someone says "help me get through my inbox", "triage these emails", "I have 200 unread messages, what needs a reply", or "turn this pile of email into a to-do list". Do NOT use for designing the ongoing system that keeps an inbox at zero - filters, folder setup, unsubscribe habits, daily routines - use inbox-zero instead.
What it loads into context
Frontmatter
129 tokensAlways available to the agent
SKILL.md
1,396 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: triage report template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore5,973 bytes
Read SKILL.md
# Email Triage
Process, don't browse. The expensive failure in email is re-reading: opening a message, feeling vague dread, closing it, and opening it again tomorrow - five touches for one decision. This skill forces exactly one touch and one decision per message, and turns the batch into a short action list with the replies already drafted.
## Inputs to collect
Before triaging, confirm - and label guesses as guesses:
1. The batch: which messages (a pasted list, a date range, "everything unread"). Default: everything unread.
2. VIP list: names whose messages surface first (boss, key clients, family). Default: infer from sender frequency and title, flagged as a guess.
3. Delegation targets: who can own what kinds of asks. Default: none - nothing gets delegated without a named owner.
4. Task system: where deferred items land (todo app, calendar, plain list). Default: a plain dated list in the output.
5. Reply register: default tone. Match the sender - brief to brief, formal to formal.
emergency-fund-planner
When your agent uses it
Sizes an emergency fund from essential expenses and household risk tier, picks the right account, and builds the funding and replenishment schedule. Use when someone asks "how much should I have in savings", "how big should my emergency fund be", "where should I keep my emergency fund", or "should I save or pay off debt first". Do NOT use for sequencing debt payoff itself - use debt-payoff-planner instead; for planned irregular expenses like car registration or gifts, those are sinking funds - use budget-builder; for the full save-vs-invest ordering - use financial-planner.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,508 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,907 bytes
Read SKILL.md
# Emergency Fund Planner
An emergency fund is not a savings account - it is insurance against being forced into high-interest debt during a crisis. The costly mistake this skill prevents is the unsized fund: either so small the first real emergency lands on a 24% APR card, or so large that years of surplus sit in cash while high-interest debt compounds. Size it to the household's actual risk, place it where it is liquid but not frictionless, and schedule both the build and the refill.
## Operating procedure
### Step 1: Gather inputs
Label guesses as guesses; refine from statements.
1. Essential monthly expenses only - the costs that cannot be paused: rent or mortgage, utilities, groceries, minimum debt payments, insurance premiums, transportation. Not total spending; a fund sized on total spending overshoots by 30-40% because discretionary spending stops in a crisis.
2. Household income structure: one earner or two, W-2 or variable/freelance, and how specialized the field is (proxy: realistic months to replace the job).
3. Current liquid savings and account type.
4. Any debt above 7-8% APR (changes the build order in Step 3).employment-contract
When your agent uses it
Reviews an employment agreement clause by clause, flags deviations from market-standard terms - non-compete scope, IP assignment breadth, cause definitions, equity mechanics - and produces a review table with recommended asks, explicitly bounded as not legal advice. Use when someone asks "can you review my offer letter", "is this non-compete normal", "what does this IP assignment clause mean", "should I push back on this severance term", or has an employment agreement in hand before signing. Do NOT use for negotiating the compensation number itself - use salary-negotiation instead - or for drafting a company's terms of service or privacy policy - use terms-of-service instead.
What it loads into context
Frontmatter
148 tokensAlways available to the agent
SKILL.md
1,790 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: review table rows
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore9,176 bytes
Read SKILL.md
# Employment Contract Review
An employment contract hides most of its risk in standard-looking boilerplate, and the clauses that hurt are the ones that only activate on the way out: termination, non-compete, IP assignment. This skill walks every key clause, flags what deviates from market norms, and frames the negotiation - so the signer knows which three things to push on instead of either signing blind or fighting every line.
This is contract literacy and negotiation framing, not legal advice. Every material concern - and any restrictive covenant the person might realistically test - goes to a licensed employment attorney in the relevant jurisdiction before signing. Enforceability of non-competes and similar covenants varies sharply by jurisdiction, and no review table substitutes for counsel. State this boundary in the output itself.
## Operating procedure
Read the contract as the party with less power. Every clause that is vague, broad, or one-sided favors the drafter - usually the employer. The job is to surface those and ask whether each is standard or aggressive.
### Step 1: Gather inputs
1. The full agreement plus every referenced exhibit (equity plan, confidentiality agreement, prior-inventions schedule, arbitration agreement). A review of the offer letter alone is incomplete - say so if that is all that exists.
2. Jurisdiction (governing-law clause and where the person will actually work).eng-status-rollup
When your agent uses it
Rolls up engineering team status into an exec-ready written update with exactly three sections - progress in outcomes, risks with likelihood and mitigation, and asks with owners and deadlines - scannable in under 90 seconds. Use when someone asks "write my weekly eng update", "summarize team status for leadership", "turn these standup notes into an exec update", or is prepping for a leadership sync. Do NOT use for updates to external or cross-functional stakeholders on a project - use stakeholder-update instead - for public incident communication - use status-page-update - or for monthly investor letters - use investor-update-writer instead.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
1,548 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template
- Worked contrast
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,771 bytes
Read SKILL.md
# Eng Status Rollup
Execs do not need a feature list. They need three answers: are we on track, what might go wrong, and what do you need from us? A rollup that buries those answers under activity logs trains leadership to skim it - and then the one week the update contains a real red flag, nobody reads it. Everything in this skill defends the 90-second scan.
## Operating procedure
### Step 1: gather inputs
- The raw material: standup notes, sprint board, incident log, whatever exists for the period since the last update.
- The last rollup sent, so claims stay consistent - a risk that silently disappears without a resolution line destroys credibility faster than the risk itself would have.
- The audience (direct exec, VP group, company-wide) and the milestone or date the team is publicly committed to.
- Anything the manager knows but hasn't written down; ask directly "what are you worried about this week?" - the real risk is usually in the answer, not the ticket system.
If a claim is unverified ("I think the migration is on schedule"), label it as unconfirmed rather than rounding up to green.engagement-reply-drafter
When your agent uses it
Drafts short, on-brand replies to public social comments and DMs, including graceful handling of praise, questions, complaints, and criticism. Use when asked to reply to a comment section or inbox - a batch of Instagram, X, TikTok, LinkedIn, YouTube, or Facebook comments or DMs - in the brand's voice. Do NOT use when answering an inbound support ticket or help-desk message - use support-ticket-reply instead.
What it loads into context
Frontmatter
114 tokensAlways available to the agent
SKILL.md
885 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,041 bytes
Read SKILL.md
# Engagement Reply Drafter
Turn a comment section or DM inbox into specific, human, on-brand replies that grow the thread without sounding like a bot.
## Workflow
1. Triage each item into one lane: praise, question, complaint/bug, criticism, troll/bad-faith, or spam/abuse. The lane sets the tone; a mismatched lane produces the wrong reply. Order the batch by urgency: complaints and questions first (aim to answer within 24 hours - past that, silence reads as ignoring), then praise and criticism. Replies inside the first 60 minutes after a post goes live also compound algorithmic reach, so prioritize fresh posts.
2. Read what the person actually said and name a concrete detail from it. Never open with a generic platitude ("Thanks so much!").
3. Draft per lane:
- Praise: acknowledge the specific thing they liked; add one natural return question only if it extends the thread without forcing it.
- Question: answer it directly first, then optionally invite more.
- Complaint/bug: empathize, own it in one line, and if it involves private details (orders, accounts), reply publicly with brief empathy then move it to DM.
- Criticism: if fair, acknowledge sincerely, take responsibility, state what you'll do, and stop - do not over-explain. If bad-faith, write a brief unbothered reply or recommend no reply.
- Spam/abuse: do not engage; recommend hide or block.env-doctor
When your agent uses it
Diagnose and fix a broken local development environment methodically - reproduce the failure, fingerprint runtime, package manager, OS and arch, check version files and PATH shadowing, lockfile drift, missing env vars (without printing secrets), port conflicts, stale caches, native-module and Apple Silicon arch mismatches, and Docker context - then apply the smallest fix and leave a reproducible setup note. Use when someone asks "it works on my teammate's machine but not mine", "npm install broke everything", "why is the wrong Node version running", "port 3000 is already in use", or "this native module fails to load". Do NOT use for bugs in application logic - use obra-superpowers-skills-systematic-debugging instead - or for planned dependency upgrades - use wshobson-dependency-upgrade instead.
What it loads into context
Frontmatter
182 tokensAlways available to the agent
SKILL.md
2,321 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked diagnosis transcript
- Thresholds
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytesSKILL.mdcore9,696 bytes
Read SKILL.md
# Environment Doctor
"Works on my machine" failures are almost always a difference in versions, paths, or state - not in code. The costly mistake this skill prevents is the shotgun fix: deleting `node_modules` and the lockfile, upgrading the runtime, reinstalling the toolchain, and reaching a state that happens to work but that nobody can explain or reproduce - while silently upgrading dozens of dependencies and leaking secrets into a pasted log along the way.
## Operating procedure
Diagnose before changing anything. Each step either finds the cause or rules out a whole class.
### Step 1: Reproduce and capture the exact failure
Run the failing command yourself and save the full output, not a summary. Note the first error, not the last - later errors are usually fallout. Ask: did this ever work on this machine? What changed since (OS update, new branch, `brew upgrade`, new terminal app)? If it works on CI or a teammate's machine, that machine is your reference.
### Step 2: Capture the environment fingerprint
error-budget-policy
When your agent uses it
Defines and applies an SLO error-budget policy that gates feature work versus reliability work based on remaining budget. Use when setting up SLOs or when high burn rate lacks a formal team response protocol.
What it loads into context
Frontmatter
58 tokensAlways available to the agent
SKILL.md
1,069 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Define the Budget
- Four Policy States
- Budget Consumption Rules
- Burn-Rate Alerting
- Governance
- Deliverable
- Quality bar
- Do / Don't
Files
SKILL.mdcore5,053 bytes
Read SKILL.md
# Error Budget Policy
An error budget without a policy is a number nobody acts on. The policy converts budget state into team behavior automatically, without requiring a manager decision each time.
## Define the Budget
Error budget = (1 - SLO target) x window duration.
Example: 99.9% SLO over 30 days = 0.1% x 30 days x 24 hours = 43.2 minutes of allowed downtime.
Track budget as a percentage remaining, not as raw minutes. 'Budget: 34% remaining' is actionable. 'Budget: 14.7 minutes remaining' is not.
## Four Policy States
error-handling
When your agent uses it
Designs typed, observable, recoverable error handling - an explicit taxonomy of retryable vs terminal failures, Result types at boundaries, single-point logging, and retry policies with real backoff numbers. Use when someone asks "should I retry this error", "how do I structure error types", "why are my logs full of duplicate stack traces", "where do I put the try/catch", or is designing failure paths for an API, job, or client. Do NOT use for building the circuit breaker component itself - use circuit-breaker-builder instead; do NOT use for handling 429s against third-party APIs - use rate-limit-handler instead; do NOT use for writing user-facing error copy - use error-message-writer instead.
What it loads into context
Frontmatter
163 tokensAlways available to the agent
SKILL.md
1,827 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: boundary with policy applied
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,086 bytes
Read SKILL.md
# Error Handling
Most production incidents are not caused by the original failure - they are caused by the handling: a retry storm that turns one slow dependency into a full outage, a swallowed exception that surfaces as silent data corruption three weeks later, or duplicate logging at five layers that buries the root cause. This skill designs error handling as a first-class part of the system: every failure classified, every class given one owner and one response.
## Operating procedure
### Step 1: Gather inputs
- The unit of work being protected (a request, a job, a batch, a UI subtree).
- Every external dependency it touches and which operations are idempotent.
- Where errors currently surface (logs, tracker, user) and who consumes each.
- Latency budget - it caps how much retrying is even allowed.
### Step 2: Build the error taxonomyerror-message-writer
When your agent uses it
Rewrites error messages to be specific, actionable, and blame-free - what happened, why it helps to know, and what to do next - with tone calibrated to severity and developer-facing detail kept separate from the user-facing text. Use when someone asks "rewrite this error message", "our app just says something went wrong", "make this validation message less hostile", or is auditing error states before launch. Do NOT use for reviewing all in-product copy beyond errors - use ux-writing-audit instead; for the copy users see during first-run setup, use onboarding-copy.
What it loads into context
Frontmatter
133 tokensAlways available to the agent
SKILL.md
837 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The three jobs of an error message
- Process
- Rules
- Tone calibration
- Structure of a message
- Examples of the rewrite
- Anti-patterns
- Output
Files
SKILL.mdcore3,556 bytes
Read SKILL.md
# Error Message Writer
You rewrite error messages. A good error message turns a dead end into a next step. The user is already frustrated; your job is to help, not to scold.
## The three jobs of an error message
1. **Say what happened** - specifically, in plain language.
2. **Say why** (if it helps) - only if it aids recovery; skip blame.
3. **Say what to do next** - the most important part. Give a clear action.
## Process
1. For each error, identify: what actually failed, what the user can do about it, and whether it's the user's action or a system fault.
2. Rewrite to be specific, human, and actionable.escalation-summary
When your agent uses it
Writes a tight escalation summary to engineering or management with a one-line headline, quantified customer impact, repro steps, prior attempts, and a single clear ask. Use when someone says "I need to escalate this ticket", "write this up for engineering", "this bug needs eng eyes", or a support issue must leave the queue for specialist action. Do NOT use for replying to the customer - use support-ticket-reply instead; do NOT use for assigning incident severity and coordinating an active outage - use sev-triage instead; do NOT use for the after-the-fact incident writeup - use postmortem-writer instead.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
1,399 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Required sections (always all five, in this order)
- Severity thresholds
- Worked artifact: copy-paste template
- Evidence package
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,529 bytes
Read SKILL.md
# Escalation Summary
Engineers and managers receive escalations all day; a well-built summary gets a response, a vague one gets queued behind everything else. The costly mistake this skill prevents is the partial escalation - sent early with "details to follow" - which burns the escalation team's first look and resets the clock on the customer. Write as if the reader has zero context and thirty seconds: if the first line does not tell them what is broken, for whom, and since when, the rest never gets read.
## Operating procedure
Order matters: qualify before gathering, gather before writing - an escalation written before the evidence exists gets bounced back for the missing pieces.
1. Qualify the escalation. Confirm the issue is genuinely outside support's scope to resolve. If a documented workaround or known fix exists, apply it instead of escalating.
2. Gather the inputs below completely. Do not start writing until everything is in hand.
3. Assign severity using the definitions below and put the label in the subject line or ticket title.
4. Write the five required sections in order, headline first. The first line is the whole escalation in miniature - spend the most editing time there.
5. Attach the evidence package.
6. Send through the designated escalation channel, never copying the customer.executive-summary
When your agent uses it
Compresses long reports, memos, and analyses into a decision-ready one-page summary (250-350 words) - bottom line up front, 3-5 evidence-backed key findings, a specific owned-and-dated recommendation, risks with mitigations, and the exact decision or resource being requested - plus a separate list of unverifiable claims to probe. Use when someone says "summarize this report for the exec team", "give me a one-pager on this analysis", "write the BLUF for this memo", or a leader needs to act on a document in 90 seconds. Do NOT use for narrating data and charts into an insight story - use data-story instead - or for recurring project status - use stakeholder-update instead.
What it loads into context
Frontmatter
166 tokensAlways available to the agent
SKILL.md
717 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core principle
- Process
- Structure (one page, in this order)
- Rules
- Insight extraction
- Recommendation discipline
- Length guidance
- Anti-patterns
- Output
Files
SKILL.mdcore3,028 bytesSKILL.mdcore3,028 bytes
Read SKILL.md
# Executive Summary
You compress reports, memos, and analyses into a one-page summary an executive can read in 90 seconds and act on.
## Core principle
Executives read to decide, not to learn. Lead with the conclusion. Everything else supports it.
## Process
1. Read the full source. Identify the single decision or question it serves.
2. Extract: the bottom line, 3-5 supporting findings, the recommendation, and the cost of inaction.
3. Draft top-down. The first sentence is the answer.
expansion-revenue
When your agent uses it
Builds a systematic expansion playbook for existing SaaS customers - a catalog of usage triggers (limit approach, new team adoption, milestone hit) with routing rules, outcome-anchored upsell and cross-sell talk tracks, ownership and comp models, and metrics led by net revenue retention. Use when someone asks "how do we grow revenue from existing accounts", "build an upsell playbook", "our NRR is too low", or wants trigger-based expansion instead of reps guessing. Do NOT use for designing tiers and value metrics from scratch - use saas-pricing instead; for stopping customers from leaving, use churn-reduction.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
786 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Why Expansion Wins
- The Two Motions
- Trigger-Based Expansion
- Talk Tracks
- Ownership and Incentives
- Metrics
- Anti-Patterns
- Deliverable
Files
SKILL.mdcore3,547 bytes
Read SKILL.md
# Expansion Revenue
In a mature SaaS business, the majority of new revenue comes from existing
customers, not new logos. Net revenue retention above 100% means you grow even
if you stop acquiring. This skill builds the systematic expansion motion.
## Why Expansion Wins
Selling more to a happy customer costs a fraction of acquiring a new one and
closes faster. Expansion revenue carries higher margin and signals product
value. A business with NRR > 120% compounds; one with NRR < 90% is a leaky
bucket no acquisition spend can fill.
## The Two Motionsexpense-policy
When your agent uses it
Drafts a clear, enforceable expense and approval policy - per-category spend limits, a dollar-tiered approval matrix, receipt and documentation rules, 30/90-day submission deadlines, and audit sampling. Use when someone says "write our expense policy", "who should approve what spend", "set reimbursement rules", or is onboarding a finance system or preparing for a compliance review. Do NOT use for building a departmental or personal budget - use budget-builder instead; do NOT use for the monthly close checklist - use month-end-close instead; do NOT use for general internal SOPs outside spend - use process-doc instead.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,566 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Concrete defaults (adjust with reason, never silently)
- Worked artifact: policy skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,486 bytes
Read SKILL.md
# Expense and Approval Policy
An expense policy that employees ignore is worse than no policy - it creates a false sense of control while spend leaks through exceptions. A good policy is short enough to read, specific enough to apply, and enforced consistently; the costly mistake this skill prevents is the prose-heavy policy nobody can apply at the moment of purchase, which converts every expense into a judgment call and every judgment call into precedent.
## Operating procedure
Order matters: limits must be calibrated to the company before the approval matrix is drawn, or the tiers will not match how spend actually flows.
1. Collect the inputs below.
2. Draft the five mandatory sections in order: what is reimbursable, spend limits by category, approval authority by amount, documentation requirements, submission deadlines. Each section is a table or short bulleted list - never prose paragraphs. An employee must be able to scan to any answer in under 30 seconds.
3. Calibrate the default numbers below to the company's size, geography, and travel intensity. Keep the structure even where the numbers move.
4. Add the enforcement section - a policy without stated consequences is a suggestion.
5. Close the catch-all: anything not listed is not reimbursable without explicit pre-approval.
6. Route the draft to finance and legal for sign-off before publication, and pair with the expense-tool configuration so approval tiers are enforced in software, not memory.experiment-tracking
When your agent uses it
Sets up disciplined ML experiment tracking - run logging schemas, artifact versioning, naming conventions, and reproducibility standards - and produces the run-record template a team actually follows. Use when someone asks "how should we track our ML experiments", "why can't we reproduce this result", "how do I set up MLflow or W&B for the team", or is bootstrapping a new ML project. Do NOT use for analyzing product A/B tests - use ab-test-analyzer instead; for monitoring deployed models use data-drift-monitor; for judging whether a trained model is good enough to ship use model-evaluation-report.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,624 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: run record template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,390 bytesSKILL.mdcore7,390 bytes
Read SKILL.md
# Experiment Tracking
An experiment no one can reproduce is a result no one can trust. Disciplined tracking is not overhead - it is the minimum viable scientific practice for ML, and the alternative is a promoted model whose provenance nobody can reconstruct when it misbehaves in production. This skill installs the logging schema, naming discipline, and promotion workflow that make every run auditable.
## Operating procedure
Follow the steps in order: the run schema (Step 2) must exist before naming conventions (Step 3) matter, and reproducibility standards (Step 4) are unenforceable until the schema captures seeds and environments.
### Step 1: Gather inputs
Collect these before touching tooling. Where the answer is unknown, record the default and label it a guess.
1. Team size and collaboration need. Default: solo or small team.
2. Existing tooling, if any (MLflow, Weights and Biases, homegrown spreadsheets).expert-interview
When your agent uses it
Designs an expert-interview package - sourcing criteria for the right experts, a funneled discussion guide with timing, open-then-probe question design, laddering, and a neutral probe kit - that extracts honest insight without leading the witness. Use when someone asks "help me write questions for an expert call", "build a discussion guide", "prep me for this expert network interview", or is planning customer-development or domain-expert conversations. Do NOT use for designing a full data-collection study with sampling and hypotheses - use primary-research instead; for synthesizing findings after the interviews are done, use interview-synthesis instead; for desk research that involves no live conversation, use deep-research instead.
What it loads into context
Frontmatter
159 tokensAlways available to the agent
SKILL.md
1,610 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: good/bad question contrast
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,521 bytes
Read SKILL.md
# Expert Interview
An hour with the right expert can replace weeks of desk research - or produce nothing but confirmation of what the interviewer already believed. The costly failure this skill prevents is the leading interview: questions that carry the hypothesis inside them, so the expert politely agrees and the team ships on evidence it manufactured itself. A good guide is a structure, not a script.
## Inputs to collect
1. **Learning objectives**: what decision the interviews inform, and the 3-5 things that must be learned. Separate "must learn" from "nice to know" so sections can be cut under time pressure.
2. **The hypothesis being tested**, if any - written down so it can be deliberately kept out of the questions.
3. **Interview length** (default 45-60 minutes) and format (call, in person).
4. **How many experts** and what mix of vantage points (practitioner, buyer, former insider, adjacent-market operator).
5. **Recording and consent constraints**, and any compliance rules (expert networks prohibit asking for confidential or MNPI material - respect this in question design).
## Operating procedure
explain-plan-reader
When your agent uses it
Read a captured EXPLAIN ANALYZE plan, name the single bottleneck node, and prescribe a targeted fix for it. Use when a specific query is slow and you have (or can capture) its execution plan from Postgres or MySQL and need to know WHICH node is burning the time and why.
What it loads into context
Frontmatter
83 tokensAlways available to the agent
SKILL.md
1,200 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytesSKILL.mdcore4,990 bytes
Read SKILL.md
# EXPLAIN Plan Reader
Turn one captured execution plan into a named bottleneck node and a specific fix. A plan is a tree executed bottom-up and inside-out; your job is to find the one node where time or rows explode, not to audit every line.
## Workflow
1. **Demand a real plan first.** Do not diagnose from query text alone. In Postgres, require `EXPLAIN (ANALYZE, BUFFERS, FORMAT TEXT)` - ANALYZE runs the query for real timings, BUFFERS shows cache-vs-disk reads. In MySQL, require `EXPLAIN ANALYZE`. Plain `EXPLAIN` shows estimates only and lies under data skew; if that is all you have, say so and ask for ANALYZE output before prescribing.
2. **Run it twice.** The first run can be cold cache. Trust the warm run's timings unless cold-cache I/O is itself the problem (confirm via BUFFERS: high `read=` vs `hit=`).
3. **Find the bottleneck node.** Postgres reports cumulative `actual time`; a node's own cost is its time minus its children's. Locate the node with the largest delta or the largest `loops` - a real bottleneck typically owns 50% or more of total runtime; if no single node does, the time is spread and the query shape (not one node) is the problem. `(actual rows=R loops=N)` means the node ran N times - multiply R×N for true rows. A Nested Loop with high `loops` driving a Seq Scan is the classic disaster. In MySQL, scan for `type=ALL` (full scan), large `rows` examined, and `Extra: Using filesort` / `Using temporary`.
4. **Confirm with the estimate gap.** Compare estimated vs actual rows on the hot node. A gap of 10x or more (e.g. estimated 1, actual 500000) means the planner is flying blind - that misestimate usually causes the bad node choice upstream and is the root cause, not the node itself. A gap under 2-3x is normal noise; do not chase it.
5. **Diagnose by node type and prescribe one fix:**
- Seq Scan on a large table filtered to few rows → the predicate is not served by an index. Name the column(s) and predicate shape; check whether the WHERE wraps the column in a function (defeats a plain index). Hand the access pattern to index-advisor for the actual index choice.
- Bad row estimate → stale or insufficient statistics. Run `ANALYZE the_table`; for skewed columns raise `default_statistics_target` (default 100) to 500-1000 for the affected column and re-ANALYZE.
- External merge / disk-spilling Sort → raise `work_mem` for the session (the Postgres default is a conservative 4MB; sorts spill the moment they exceed it), or provide pre-sorted input via an index ordering.
- Hash Join with `Batches > 1` (spilling to disk) → raise `work_mem`.fact-checker
When your agent uses it
Verifies discrete factual claims against multiple independent sources and returns a calibrated verdict - Verified, Likely true, Likely false, Unverifiable, or False - with confidence and citations. Use when someone asks "is this true", "fact-check this claim", "did X really happen", "verify this statistic before we publish", or pastes a quote, number, or viral post to check. Do NOT use for tracing a citation chain back to its primary origin and appraising study methodology - use claims-verifier instead; for a broad open-ended question that needs a full research brief, use deep-research.
What it loads into context
Frontmatter
140 tokensAlways available to the agent
SKILL.md
1,219 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Verdict scale and thresholds
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytesSKILL.mdcore5,996 bytes
Read SKILL.md
# Fact Checker
Treat every claim as unproven until corroborated. The job is calibrated truth, not a verdict for its own sake - and the costly failure this skill prevents is the confident verdict built on correlated sources: three outlets repeating one wire story feels like consensus and is actually a single point of failure. "I can't verify this" is a valid, honest answer; a fabricated citation is never acceptable.
## Inputs to collect
Gather these before checking anything. If the user cannot supply one, use the default and label the assumption.
1. The exact claim, verbatim. Paraphrase later; anchor on the original wording first.
2. Where it appeared and what is at stake (publishing it, acting on it, arguing against it). Stakes set the sourcing bar - anything being published gets the full two-independent-primary-sources standard.
3. Depth and deadline. Default: full check.
4. Any sources the user already has. Treat these as leads, not evidence, until appraised.
## Operating procedurefeature-flags
When your agent uses it
Designs feature-flag systems and lifecycle policy - flag taxonomy (release, ops/kill-switch, experiment, permission), percentage rollouts with stable bucketing, safe defaults, and mandatory removal deadlines so flags don't rot into dead code. Use when someone asks "how should I roll this out behind a flag", "kill switch vs feature flag", "how do I do a percentage rollout", "we have hundreds of stale flags", or is wiring up LaunchDarkly or a homegrown flag service. Do NOT use for designing the A/B test analysis itself - use ab-test-analyzer instead - and do NOT use for CI/CD pipeline gating - use github-actions instead.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,629 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: flag lifecycle record
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,271 bytes
Read SKILL.md
# Feature Flags
Flags decouple deploy from release: code ships dark and turns on safely, in stages, with an instant off switch. The costly mistake this skill prevents is flag debt - every flag is a fork in your codebase that doubles the test matrix, and a flag without an owner and a removal deadline set at creation becomes permanent dead code that someone flips three years later during an incident with no idea what it does.
## Operating procedure
### Step 1: Gather inputs
Before creating any flag, collect:
1. Purpose - which of the four types below it is. If the answer is "two of them", that is two flags.
2. Blast radius if the new path misbehaves (cosmetic / degraded / data-corrupting / revenue-affecting).
3. Bucketing key - usually user id; sometimes account id so a whole team sees one variant.
4. Owner (a person, not a team) and the removal condition.feature-store-design
When your agent uses it
Designs a reusable, leakage-safe feature store - entity and naming contracts, point-in-time correct training joins, feature versioning, online/offline consistency with staleness SLOs, and governance. Use when someone asks "should we build a feature store", "how do I share features across models", "our model trains great but serves garbage", "how do I version a feature definition", or is diagnosing training-serving skew. Do NOT use for designing the feature transformations themselves - use ml-feature-engineering instead; for monitoring feature and prediction distributions after deployment - use data-drift-monitor instead; for general relational schema design - use database-schema instead.
What it loads into context
Frontmatter
148 tokensAlways available to the agent
SKILL.md
1,774 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: the join that leaks vs the join that does not
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,396 bytesSKILL.mdcore8,396 bytesSKILL.mdcore8,396 bytesSKILL.mdcore8,396 bytes
Read SKILL.md
# Feature Store Design
Feature stores exist to solve three problems at once: reuse across models, point-in-time correctness during training, and consistency between training and serving. A design that solves only one is incomplete - and the failure mode is expensive: silent label leakage that inflates offline metrics, or training-serving skew that makes a model score 0.85 offline and behave randomly in production.
## Operating procedure
### Step 1: Decide whether a feature store is warranted
Collect: number of models in production or planned, how many features they would share, and whether any model serves online. Rule of thumb: below 2-3 models sharing features, a feature store is premature infrastructure - a documented feature pipeline with a catalog (see ml-feature-engineering) does the job. Build or adopt one when features are shared across 3+ models, when online serving needs the same values training used, or when point-in-time bugs have already burned the team. Label the decision and its inputs; revisit when the model count doubles.
### Step 2: Fix entity and naming contracts
Consistent naming is load-bearing - it is the API contract for every downstream model.
feedback-writer
When your agent uses it
Drafts specific, observable, kind workplace feedback using the SBI model - Situation, Behavior, Impact - for both praise and correction. Use when someone asks "help me give feedback to a teammate", "how do I tell my report they keep missing deadlines", "make this feedback less vague", or wants to reinforce a good behavior specifically. Do NOT use for formal performance review documents - use performance-review-writer instead; for feedback on student work, use student-feedback-writer; for mediating an active interpersonal dispute, use conflict-resolution.
What it loads into context
Frontmatter
124 tokensAlways available to the agent
SKILL.md
1,376 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The SBI model
- Operating procedure
- Worked contrast: the same feedback, bad and good
- Positive SBI example
- Template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,557 bytes
Read SKILL.md
# Feedback Writer
Vague feedback ("be more proactive") is useless and stressful: the receiver cannot act on it and cannot defend against it. The SBI model - Situation, Behavior, Impact - makes feedback concrete, actionable, and defensible, and it works identically for praise and correction. The costly failure this prevents is the slow one: months of unspoken friction detonating in a review the person never saw coming.
## The SBI model
- Situation - when and where it happened. Anchors the feedback in one real moment.
- Behavior - what the person observably did or said. No interpretation, no labels, no motives.
- Impact - the effect on you, the team, or the work. This is the "why it matters."
- Optional fourth step for growth, Intent/Next: ask about intent and agree on a path forward. Ask before prescribing - you may be missing context.
## Operating procedure
### Step 1: Gather the inputsfiction-scene-writer
When your agent uses it
Writes individual fiction scenes with goal-conflict-turn structure, sensory grounding, interiority, and subtext-driven dialogue - then notes the turn accomplished and one craft choice so the writer can learn from it. Use when someone asks "write this scene for my novel", "my dialogue feels flat", "make this moment more immersive", or has plot beats that need dramatizing on the page. Do NOT use for plotting a whole novel's arc - use story-structure-architect instead; for revising a completed manuscript, use manuscript-reviser.
What it loads into context
Frontmatter
129 tokensAlways available to the agent
SKILL.md
747 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- The anatomy of a scene
- Craft rules
- Pacing
- Dialogue
- Anti-patterns
- Output
Files
SKILL.mdcore3,396 bytes
Read SKILL.md
# Fiction Scene Writer
You write individual fiction scenes that pull the reader into a moment. A scene is not a summary; it happens in real time, in a place, through a character who wants something.
## Process
1. Gather: the POV character and what they want in this scene, the other characters, the setting, the emotional turn the scene must accomplish, and any plot beats it must hit.
2. Find the scene's tension - what's at stake right now, even in a quiet moment.
3. Write the scene, then revise for subtext and sensory grounding.
## The anatomy of a scene
- **Goal**: the POV character wants something entering the scene.
- **Conflict**: something resists - a person, the environment, themselves.finance-fx
When your agent uses it
Use when a task needs live or historical money data - "convert USD to EUR", "current/past exchange rate", "FX rate on this date / over this range", or "current price of Bitcoin/Ethereum, market cap, 24h change". Frankfurter (ECB reference rates, no key) is the FX default; CoinGecko's free keyless tier covers crypto. Do NOT use for stock quotes or equities - no keyless stock API survives verification, say so instead of guessing; do NOT use for country economic indicators like GDP or inflation series - use government-open-data instead; if the request is a vague "I need live data", route through public-data-api-picker.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,370 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Ranked APIs
- Procedure
- URL templates + real response shapes
- Deliverable
- Do NOT use - reach for X instead
- Quality bar
Files
SKILL.mdcore5,191 bytesSKILL.mdcore5,191 bytes
Read SKILL.md
# Finance & FX Data
Fetch exchange rates and crypto prices with keyless calls. The mistake this prevents: reaching for exchangerate.host, Fixer, or Alpha Vantage from training memory - all key-walled now - or hand-waving a stale rate from training data into a real calculation.
## Ranked APIs
1. **Frankfurter** (api.frankfurter.dev) - DEFAULT for fiat FX. European Central Bank reference rates, ~30 major currencies, history back to 1999, time series, no key, no rate limit worth worrying about.
2. **CoinGecko free tier** (api.coingecko.com) - crypto prices, market caps, 24h change for thousands of coins. No key, but a tight limit: roughly 5-15 calls/min from an anonymous IP. Batch ids into one call; on 429, back off 60s.
## Procedure
1. **Fiat or crypto?** Fiat → Frankfurter. Crypto → CoinGecko. Crypto→fiat conversion is one CoinGecko call (`vs_currencies=eur`), not two hops.
2. **Date the rate.** ECB reference rates update ~16:00 CET on business days; the response's `date` field tells you which day you actually got. Weekend/holiday requests return the previous business day - quote that date, don't present it as today's.
3. **Use CoinGecko ids, not tickers.** The `ids` param wants `bitcoin`, `ethereum`, `solana` - not BTC/ETH. For an unfamiliar coin, resolve once via `/api/v3/search?query={name}` → `coins[0].id`.financial-planner
When your agent uses it
Builds a complete personal financial plan - a prioritized order of operations across budgeting, emergency savings, debt payoff, and goals - from the user's actual numbers, then routes each piece to the right specialist skill. Use when someone asks "help me get my finances in order", "where should my money go first", "should I pay off debt or save", "am I doing okay financially", or "build me a financial plan". Do NOT use for constructing the monthly budget itself - use budget-builder instead; for sequencing multiple debts - use debt-payoff-planner; for sizing the cash cushion - use emergency-fund-planner; for investing fundamentals - use investment-basics; for retirement readiness math - use retirement-projection; for lowering a tax bill - use tax-optimization; for evaluating one large purchase - use big-purchase-decision.
What it loads into context
Frontmatter
186 tokensAlways available to the agent
SKILL.md
1,691 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,309 bytes
Read SKILL.md
# Personal Financial Planner
This is the umbrella skill: it turns a messy money situation into one prioritized plan and dispatches the detailed work to specialist skills. The costly mistake it prevents is optimizing the wrong piece first - contributing to a brokerage account while carrying a 24% APR credit card, or attacking debt with zero cash buffer so the first surprise expense goes straight back on the card. Order of operations is the whole game; be clear, practical, and non-judgmental about wherever the user is starting from.
## Operating procedure
### Step 1: Gather inputs
Collect these before recommending anything. If a number is a guess, label it a guess and proceed; refine later.
1. Monthly take-home income (net, all sources). For variable income, use the average of the last three months.
2. Essential monthly expenses: rent or mortgage, utilities, groceries, insurance, minimum debt payments, transportation.
3. Every debt: creditor, balance, APR, minimum payment.
4. Current cash savings and where it sits.financial-statement-builder
When your agent uses it
Builds and interprets the income statement, balance sheet, and cash flow statement as one linked system - construction order, the three mechanical ties between statements, and the review checks that catch errors. Use when someone asks "build me a three-statement model", "why doesn't my balance sheet balance", "how does net income flow into the cash flow statement", "check this financial package for errors", or is explaining statements to stakeholders. Do NOT use for spreadsheet layout and formula hygiene - use spreadsheet-model-builder instead - for forward-looking 13-week or 12-month cash projections - use cash-flow-forecast instead - or for the monthly process that produces the trial balance - use month-end-close instead.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
2,178 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: how the three statements tie
- Review thresholds
- Red flags on review
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore10,371 bytesSKILL.mdcore10,371 bytes
Read SKILL.md
# Financial Statement Builder
The three core statements are one system, not three documents: an error in one surfaces in another, and a package whose statements do not tie is wrong somewhere even if each statement looks plausible alone. The costly mistake this skill prevents is shipping a package where the balance sheet is forced to balance with a plug - the plug hides a real error that compounds every period until someone has to unwind months of statements.
## Operating procedure
### Step 1: Gather inputs
1. The trial balance or account-level data for the period (if the books are not closed, run month-end-close first).
2. Opening balance sheet - the prior period's closing balance sheet. Without it, the cash flow statement and retained earnings cannot tie; if it does not exist, state that the first period's linkages cannot be verified.
3. Accounting basis (accrual assumed; flag if cash-basis) and the effective tax rate.
4. Dividends or distributions declared in the period.
5. Any non-cash items: depreciation, amortization, stock-based compensation.
fitness-program
When your agent uses it
Use when someone asks to build a HIIT plan, interval-training block, metabolic conditioning or metcon, circuit training, work-capacity training, conditioning for fat loss, or high-intensity "get in shape" cardio (not easy steady-state). Builds a metabolic-conditioning block - interval and circuit work using EMOM, AMRAP, Tabata, and intervals - with format selection, work-to-rest ratios, weekly structure, and progression. General fitness guidance, not medical advice. This skill owns high-intensity interval conditioning only. Do NOT use to design a resistance/strength program for getting stronger or bigger (use Strength Training Plan), a steady-state aerobic-base or Zone 2 block (use Zone 2 Cardio Plan), or a mobility/flexibility routine (use Mobility Routine).
What it loads into context
Frontmatter
179 tokensAlways available to the agent
SKILL.md
1,506 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Confirm Scope Before Programming
- Gate Intensity to Readiness
- Select the Format to the Goal
- Choose Movements That Survive Fatigue
- Build the Weekly Structure
- Progress Without Just Adding Intensity
- Deliverable
- Quality Bar
- What NOT to Do
Files
SKILL.mdcore6,849 bytes
Read SKILL.md
# Conditioning & HIIT Program
This skill builds a high-intensity metabolic-conditioning block: interval work, circuits, and timed formats (EMOM, AMRAP, Tabata, fixed intervals) aimed at work capacity and fat-loss conditioning. The default is a 4-week block, 3 conditioning sessions per week, repeatable. This is general fitness guidance, not medical advice - anyone with cardiac risk factors, on heart-rate-affecting medication, pregnant, or returning from illness or injury should clear high-intensity training with a physician first.
## Confirm Scope Before Programming
Establish four things first: the goal (fat-loss conditioning, work capacity for a sport, or general high-intensity fitness), training age and current conditioning level, available equipment and space, and days per week. If the request is actually about getting stronger or bigger, building an aerobic base at easy effort, or improving flexibility, stop and hand off - strength belongs to Strength Training Plan, steady-state and Zone 2 to Zone 2 Cardio Plan, mobility to Mobility Routine. This skill is for high-intensity intermittent work, not easy continuous cardio.
## Gate Intensity to Readiness
True HIIT means near-maximal efforts; it is potent and easy to overdo. Beginners (under ~3 months of consistent training) start with longer rest and circuit formats, not all-out intervals - work-to-rest around 1:3 or 1:4. Intermediates tolerate 1:1 to 1:2. Only well-conditioned trainees should run 2:1 or true Tabata (20s on / 10s off, all-out). Cap genuine high-intensity work at 2-3 sessions per week regardless of level; the adaptation comes from recovering between hard efforts, not from doing them daily.
## Select the Format to the Goal
flaky-test-detangler
When your agent uses it
Root-causes intermittently failing tests and eliminates the hidden dependency at its source instead of retrying around it. Use when a test passes locally but fails in CI, goes green on a CI re-run, fails roughly one run in ten, or is already tagged "flaky." Do NOT use when the task is to design the fake or stub that replaces a real dependency - use mock-stub-designer instead.
What it loads into context
Frontmatter
104 tokensAlways available to the agent
SKILL.md
1,179 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Worked example
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,195 bytesSKILL.mdcore5,195 bytes
Read SKILL.md
# Flaky Test Detangler
A flaky test is not random - it is a hidden dependency on something you did not control (order, timing, shared state, a real clock, or a real network). Your job is to name that dependency and remove it, never to add a retry.
## Workflow
1. **Reproduce before theorizing.** Never diagnose from a single failure. Run the suspect test 100x in isolation, then 100x inside the full suite. Failing only in the suite means cross-test contamination; failing alone means the test owns the flake.
2. **Randomize order and capture the seed.** Run with randomized order (Jest `--shuffle`, RSpec `--order random`, pytest-randomly) and record the failing seed plus the order so the failure is reproducible. This seed is your gating evidence - do not accept any fix until it makes the captured seed pass deterministically.
3. **Kill fixed-time waits.** The top cause is waiting on a sleep instead of a condition. Replace `setTimeout`/`sleep` with explicit waits on observable state: Playwright auto-waiting locators and `expect.poll`, Cypress retry-ability, awaitility on the JVM. Never assert immediately after firing an async action.
4. **Isolate shared mutable state.** Tests that pass alone but fail together share a singleton, a module-level cache, a seeded row, an unrolled-back transaction, an env var, or the system clock. Make each test set up and tear down its own world: wrap DB tests in a per-test transaction that rolls back, or truncate between tests.
5. **Freeze the clock and cut the network.** Pin time with fake timers (`jest.useFakeTimers`, sinon, freezegun) so midnight and DST cannot break a test. Replace live outbound HTTP with a controlled stub - a test hitting a real endpoint is flaky by definition. When the stub or fixture itself needs designing, hand that off to mock-stub-designer; here you only require that the dependency is no longer real.
6. **Remove order assumptions.** Never assume insertion order from a hash, set, or unordered query - sort before asserting.
7. **Quarantine only what you cannot yet fix.** Tag a confirmed-flaky test as skipped-but-still-running (non-blocking) with a linked ticket, an owner, and a deadline, so you keep collecting failure data. Quarantine is a holding cell, not a verdict.
8. **Stop at the root cause.** Close the case when you can name the dependency and the captured seed passes deterministically. If 500x of looping yields no repro and the test guards critical behavior, ship structured logging and the failing seed to CI artifacts instead of deleting it. Delete a test only when it asserts nothing meaningful or duplicates coverage.
flutter-widget-architect
When your agent uses it
Architects Flutter widget trees and Riverpod or Bloc state so rebuilds are scoped, build() stays pure, and const widgets skip recomposition. Use when someone asks "why is my Flutter screen janky", "should I use Riverpod or Bloc", "my whole page rebuilds when one field changes", or is building or refactoring Flutter screens, wiring StatelessWidget/StatefulWidget, Consumer, or BlocBuilder. Do NOT use for Android Jetpack Compose UI - use jetpack-compose-builder instead; do NOT use for React Native - use react-native-pro instead.
What it loads into context
Frontmatter
133 tokensAlways available to the agent
SKILL.md
1,553 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,885 bytes
Read SKILL.md
# Flutter Widget Architect
Structure Flutter widget trees and state so rebuilds start and stop exactly where intended and build() is a pure, cheap function of state. The costly failure this prevents is the rebuild storm: one text field changes, the entire screen re-runs build(), the frame budget blows past 16ms (60fps) or 8ms (120fps), and the user sees jank that profiling later traces to architecture, not to any single slow line.
## Operating procedure
Order matters: decompose before optimizing subscriptions, because `.select` on a monolithic build() buys nothing.
### Step 1: Gather inputs
Before restructuring a screen, establish:
1. The screen or widget in question and where jank or excessive rebuilds appear (default: run DevTools rebuild tracking or `debugPrintRebuildDirtyWidgets` first and capture which widgets rebuild).
2. Existing state management - Riverpod, Bloc, provider, setState, or a mix (a mix inside one feature is itself the finding).formatted-report-writer
When your agent uses it
Produces long-form business and consulting reports with a standalone executive summary, three-level heading hierarchy, and skimmability rules that let a busy reader navigate in under two minutes. Use when someone asks "write up this analysis as a report", "structure my strategy brief", "turn these findings into a client deliverable", or "make this document skimmable for executives". Do NOT use for presenting user-research findings to a product team - use research-readout instead; for a one-paragraph-to-one-page condensation of an existing document, use executive-summary; for slide-based delivery, use slide-deck-builder; for academic papers, use academic-essay.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,296 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast: executive summary opening
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,241 bytes
Read SKILL.md
# Formatted Report Writer
A business report has two audiences: the executive who reads only the summary, and the analyst who reads everything. The costly failure this skill prevents is the report that serves neither - findings buried in chronology, a summary that teases instead of concludes, and a busy reader who gives up before reaching the recommendation they were supposed to approve.
## Operating procedure
### Step 1: Gather inputs - do not draft until these are settled
Answer in one sentence each; if the requester cannot, propose an answer and label it a guess:
1. What decision or action does this report enable?
2. Who is the primary reader, and what do they already know?
3. What must the reader believe by the end?
4. Register: formal (board, regulatory) or plain (internal ops)? Default: plain.fpa-model
When your agent uses it
Builds a driver-based FP&A operating model linking business inputs to P&L, balance sheet, and cash flow outputs. Use when building an annual plan, preparing investor materials, running scenario analysis, or stress-testing the business.
What it loads into context
Frontmatter
64 tokensAlways available to the agent
SKILL.md
1,131 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Model Architecture
- Revenue Build
- Expense Build
- Scenario Framework
- Model Hygiene
- Key Outputs to Surface
- Do NOT
- Quality bar
Files
SKILL.mdcore5,409 bytes
Read SKILL.md
# FP&A Operating Model
A driver-based model is built on the belief that financial outputs are the result of measurable business actions. Revenue is not a number pulled from a goal - it is a function of pipeline, conversion rate, average contract value, and sales capacity. Every assumption should be traceable to a business driver, not a wish.
## Model Architecture
Structure the model in four layers. The Assumptions layer holds every input: growth rates, hiring plan, pricing, margins, DSO, and macro variables. The Revenue layer translates demand drivers into recognized revenue. The Expense layer translates the headcount plan and operational drivers into opex and COGS line items. The Output layer assembles the three financial statements automatically from the above. Never hardcode a number in the output layer - it must flow from the assumptions.
## Revenue Build
For SaaS or subscription: start with beginning ARR, add new ARR from new logos (pipeline times close rate times ACV), add expansion ARR, subtract churn ARR (beginning ARR times churn rate), to get ending ARR. Monthly revenue equals ending ARR divided by 12. For transaction businesses: units sold times ASP, by segment. For services: billable headcount times utilization rate times bill rate. Make the revenue build auditable - one row per driver. Sanity-check the drivers against benchmarks: a modeled close rate above 25-30% on cold pipeline, or gross churn assumptions below 5% annually for an SMB product, deserve a written justification because they are outside typical ranges.
## Expense Build
framework-upgrader
When your agent uses it
Drives a major-version framework bump as a sequence of small, reversible, CI-gated PRs using official codemods and changelog diffing while keeping the app green. Use when bumping React, Rails, Spring Boot, Angular, or any framework across a major version with breaking API/config changes; do NOT use for language or runtime version jumps (Python 2 to 3, Node majors, Java LTS) - use language-version-migrator instead.
What it loads into context
Frontmatter
112 tokensAlways available to the agent
SKILL.md
1,055 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,825 bytes
Read SKILL.md
# Framework Upgrader
Treat a major-version framework upgrade as a sequence of small, mergeable, reversible PRs against a green main, driven by the framework's official migration path. A single long-lived upgrade branch is the failure mode.
## Workflow
1. **Diff the changelog into a concrete edit list.** Open the framework's own upgrade guide (React upgrade guide, Rails `rails app:update`, Spring Boot migration notes, Angular `ng update`). Grep your codebase for the removed/renamed APIs, deprecated config keys, and changed defaults the guide names. Produce a checklist of specific edits with file locations - never a vague "upgrade React." Split each item into required-now vs. deferrable-deprecation.
2. **Get green on the current major first.** Confirm the suite passes and clear every deprecation warning on the current version in their own PRs, since most frameworks promote those warnings to errors in the next major. This is the highest-leverage prep step: it shrinks the actual jump to near zero. Never skip an intermediate major - walk one major at a time (4→5→6, not 4→6), because guides and codemods assume single-step jumps.
3. **Run the official codemods per area, in isolated commits.** Apply the provided transforms (react-codemod / `npx codemod`, Angular `ng update` schematics, OpenRewrite recipes). One area per commit so each diff is reviewable. Write custom jscodeshift/AST transforms for repeated project-specific patterns the official codemods miss.
4. **Review every hunk; codemods are not authoritative.** Read each generated change. Codemods skip dynamic/indirect usages and occasionally rewrite incorrectly. Hand-fix what they missed before the PR lands.
5. **Run old and new side by side where the toolchain allows.** Use the framework's coexistence path (React gradual adoption, Rails dual-boot with a second Gemfile, Angular compatibility modes). Land prep and codemod PRs to main continuously; keep each independently revertible and CI-gated. Keep hand-written diffs under roughly 400 changed lines per PR; uniform mechanical codemod output can run larger, but flag it as machine-generated so reviewers spot-check instead of line-reading.
6. **Verify beyond compilation.** Run the full suite, smoke-test critical flows, and check bundle size and startup time - treat a regression beyond ~5% on either as a stop-and-investigate signal. Watch error rates on a canary for at least 24 hours (one full traffic cycle) before rolling wide. New majors change defaults (rendering, serialization, timezone, strictness) that pass type checks but alter behavior - diff observable output where you can.
7. **Stop and re-scope a blocked dependency.** If a dependency has no version compatible with the target and no maintained fork, pause. You are now also migrating off that library, which is a separate project - flag it, scope it, and sequence it on its own.
## Deliverablefreelance-positioning
When your agent uses it
Selects a defensible freelance niche and writes a positioning statement using the vertical x problem x outcome procedure, then aligns portfolio proof to it. Use when a freelancer or consultant says "I take any work that pays", "how do I niche down", "what should I specialize in", "write my positioning statement", "my portfolio is all over the place", or "clients see me as a pair of hands, not an expert". Do NOT use for positioning a product or company in a market category - use positioning-statement instead.
What it loads into context
Frontmatter
115 tokensAlways available to the agent
SKILL.md
1,564 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Concrete thresholds
- Deliverable
- Do NOT
- Quality bar
- Next in the pack
Files
SKILL.mdcore7,343 bytes
Read SKILL.md
# Freelance Positioning
A generalist freelancer competes on price against everyone; a positioned freelancer competes on relevance against almost no one. The costly mistake this skill prevents is staying broad out of fear: "I don't want to turn away work" keeps rates pinned at commodity levels because no client pays a premium for interchangeable. Install this skill first in the Freelancer / Consultant OS pack - the niche chosen here is the input to productized-service-designer, and the pricing in retainer-pricing-calculator only holds up if the positioning supports it.
Worked example throughout the pack: Maya, a freelance brand designer, currently earning $6,000 a month on hourly work at $75/hr, moving to productized retainers.
## Operating procedure
Follow the steps in order. The niche grid must come before the statement, and the statement before the portfolio audit, because each step is the raw material for the next.
### Step 1: Collect the inputs
Gather these from the freelancer. If an answer is a guess, label it a guess and move on.
fun-content
When your agent uses it
Use when a task needs light entertainment content from a live API - "build a quiz/trivia game", "random dog picture", "cat fact", "advice of the day", "random fun fact" - for a demo, bot, party feature, or filler content slot. Open Trivia DB is the quiz default (its session-token mechanics are documented here); catfact.ninja, random.dog, Advice Slip, and Useless Facts cover the rest, all keyless. Do NOT use for placeholder users/products/images in a prototype - use test-placeholder-data instead; do NOT use for dictionary or encyclopedia lookups - use language-reference instead; if the request is a vague "I need live data", route through public-data-api-picker.
What it loads into context
Frontmatter
167 tokensAlways available to the agent
SKILL.md
1,385 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Ranked APIs
- Procedure
- URL templates + real response shapes
- Deliverable
- Do NOT use - reach for X instead
- Quality bar
Files
SKILL.mdcore5,680 bytesSKILL.mdcore5,680 bytes
Read SKILL.md
# Fun Content APIs
Pull trivia, animal images, and one-liner facts from verified keyless endpoints. The mistakes this prevents: shipping a quiz that repeats questions because Open Trivia DB's token mechanics were skipped, printing raw HTML entities in questions, and citing one of this category's chronically-dead quote APIs from training memory.
## Ranked APIs
1. **Open Trivia DB** (opentdb.com) - DEFAULT for quiz/trivia: 4,000+ questions, categories, difficulties. No key, but has session-token and encoding mechanics (below) that naive integrations get wrong.
2. **catfact.ninja** - random cat facts. **random.dog** - random dog images (the better-known Dog CEO API at dog.ceo is returning HTTP 520s, verified - see Do NOT).
3. **Advice Slip** (api.adviceslip.com) and **Useless Facts** (uselessfacts.jsph.pl) - one-liner advice and fun facts.
## Procedure
1. **Trivia beyond one throwaway call → get a session token first.** The token makes the API track served questions so users never see repeats. Rate limit: **1 request per IP per 5 seconds** - batch up to 50 questions per call instead of calling per-question.
2. **Decode the questions.** Default responses are HTML-encoded (`"`, `'`). Either decode entities client-side or request `&encode=url3986` and URL-decode - do one of them, or apostrophes render as garbage.fundraise-closing-mechanics
When your agent uses it
Drives a fundraise from signed term sheet to wired money - a closing checklist covering confirmatory diligence with same-day response discipline, the definitive-docs list (stock purchase agreement, charter, investor rights, board consents), closing conditions with owners and dates, re-trade defense, multi-party sequencing behind the lead, and post-wire actions like share issuance and the first investor update. Use when a founder asks "what happens after the term sheet", "how long does closing take", "the investor is dragging on confirmatory diligence", or "when does the money actually arrive". Do NOT use for negotiating the term sheet itself - use term-sheet-negotiation instead - or for assembling the diligence materials - use data-room-builder instead.
What it loads into context
Frontmatter
157 tokensAlways available to the agent
SKILL.md
825 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The path from signed term sheet to cash
- Keep momentum through the stretch
- Watch for re-trades
- After the wire
- Anti-patterns
- Deliverable
Files
SKILL.mdcore3,843 bytes
Read SKILL.md
# Fundraise Closing Mechanics
A term sheet is a non-binding promise; rounds still die between signing and
funding. The home stretch is where founder attention drains, diligence surprises
surface, and momentum quietly leaks. This skill drives the deal to wired funds.
## When to use this skill
Use it the moment a term sheet is signed (and [[term-sheet-negotiation]] is
done), running against the data room from [[data-room-builder]].
## The path from signed term sheet to cash
1. **Confirmatory diligence** - the investor verifies what you claimed: metrics,fundraise-readiness-audit
When your agent uses it
Runs the audit an investor will run before a founder starts pitching - scoring metrics, story, team, and legal/financial hygiene red/yellow/green and returning a go, fix-first, or wait verdict with the milestones that turn each red gate green. Use when a founder asks "audit my fundraise readiness before I pitch", "run the audit an investor would run on my startup", "score my metrics, story, and team before I raise", "am I ready to start fundraising", or wants a fundraise checklist before any outreach at any stage. Do NOT use for Series A-specific metric benchmarks (ARR bar, NRR, payback) and building the Series A data room - use series-a-readiness. Do NOT use for choosing the stage, amount, and instrument of the raise - use fundraising-stage-selector. Do NOT use for writing the pitch story itself - use fundraising-narrative.
What it loads into context
Frontmatter
205 tokensAlways available to the agent
SKILL.md
1,845 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Scorecard template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,858 bytes
Read SKILL.md
# Fundraise Readiness Audit
A raise that starts before the company is ready burns the best investors first - and a founder gets exactly one first meeting with each. This skill runs the audit a sharp investor will run, at any stage, so the founder finds the holes before the investor does. It is stage-agnostic: it decides go versus wait, not whether the numbers clear a specific round's benchmark - that is series-a-readiness territory when the round is an A.
## Operating procedure
Run it after fundraising-stage-selector sets the envelope (stage, amount, instrument) and before any outreach. Re-run it the week before going live to confirm nothing regressed.
### Step 1: gather inputs
- The stage envelope: round, target amount, and the milestone the money buys (from fundraising-stage-selector).
- Current metrics: the hero metric with its denominator, growth, retention or NRR, burn, runway, CAC/payback - with sources. Label any unreconciled or estimated figure as a guess.
- The narrative draft (from fundraising-narrative) and the deck if one exists.
- Cap table, option grants, prior SAFEs/notes, IP assignments, incorporation docs - or an honest statement of what is missing.fundraise-team-hiring
When your agent uses it
Use when a founder is shaping the team for a raise or hiring against the round. Triggers on "team slide", "what hires do investors want", "how much equity for an early hire", "key hire before fundraising", "is my team fundable", "hiring plan for the round". Connects the team story, the gap-filling hires, and the equity to grant.
What it loads into context
Frontmatter
89 tokensAlways available to the agent
SKILL.md
1,284 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The team story investors fund
- The hires that de-risk a round
- Equity for early hires (rough bands, calibrate to stage and seniority)
- Worked example: seed-stage devtools company
- Anti-patterns
- Deliverable
- Quality bar
Files
SKILL.mdcore5,187 bytes
Read SKILL.md
# Fundraise Team & Hiring
Investors fund a team's ability to learn faster than the market changes. "Why
this team wins this bet" must be answerable without hand-waving - and the gaps
must be named with a credible plan, not hidden. This skill builds both the story
and the hiring plan behind it.
## When to use this skill
Use it while building the team slide in [[pitch-deck-builder]], when sizing the
option pool in [[cap-table-manager]], and when use-of-funds in the ask is mostly
"key hires."
## The team story investors fundfundraising-narrative
When your agent uses it
Builds the five-beat narrative spine - origin, insight, wedge, traction, vision - that answers why now, why you, and why this before any deck exists. Use when a founder says "help me craft my fundraising story", "investors don't get what we do", "what's our narrative for this round", or has metrics and a product but no story that survives a partner meeting. Do NOT use for turning the story into slides - use pitch-deck-builder instead; do NOT use for deciding stage, amount, or valuation - use fundraising-stage-selector instead.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
1,522 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- The five-beat arc
- Tightening rules
- Stage calibration
- Worked artifact: narrative memo template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,903 bytes
Read SKILL.md
# Fundraising Narrative
Investors back conviction before they back numbers, and conviction comes from a story with a spine. The costly mistake this skill prevents is building the deck first: founders who start in slides end up with twenty loosely related pages and no answer to the three questions every partner meeting silently asks - why now, why you, why this. Build the narrative memo first; the deck is rendered from it, never the reverse.
## Operating procedure
Order matters: raw material before drafting, drafting before tightening, tightening before any slide exists.
1. Gather the inputs below. Do not draft on assumptions about traction or stage.
2. Interview the founder for raw material. The single best question is "What did you believe a year ago that you no longer believe?" - the answer is usually the Insight. For Origin, ask "Why did this problem choose you?" and push past the resume answer to the earned secret.
3. Draft the arc as six spoken sentences first - one per beat, plus one for the ask. Read it aloud. If it does not survive being spoken, it will not survive a partner meeting.
4. Test the draft against the three questions. Why now must live in Insight, why you in Origin, why this in Wedge + Traction. A beat that answers none of the three gets rewritten.
5. Apply the tightening rules below to every sentence.
6. Calibrate the weighting to stage.fundraising-stage-selector
When your agent uses it
Use when a founder is deciding whether and how to raise. Triggers on "am I pre-seed or seed", "how much should I raise", "what stage am I", "what valuation can I get", "should I raise now or wait", "is my round too big". Sets the round, amount, and valuation envelope that every later step depends on.
What it loads into context
Frontmatter
83 tokensAlways available to the agent
SKILL.md
836 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- Read the stage off evidence, not ambition
- Size the raise from the milestone, not the market
- Valuation envelope
- Should you raise at all?
- Anti-patterns
- Deliverable
Files
SKILL.mdcore3,414 bytes
Read SKILL.md
# Fundraising Stage Selector
Every later step - deck, targeting, terms - is wrong if the stage is wrong. The
most common founder error is raising at the stage they wish they were at, not
the one their evidence supports. This skill fixes the stage first.
## When to use this skill
Reach for it before writing a deck or emailing an investor, whenever the
founder asks how much to raise, what valuation is realistic, or whether to raise
at all yet. It produces the envelope; everything downstream is calibrated to it.
## Read the stage off evidence, not ambition
funnel-analysis
When your agent uses it
Builds conversion funnels from raw event data with drop-off attribution, segment comparison, and significance testing, and delivers a filled funnel table with a diagnosed bottleneck and next action. Use when someone asks "where are users dropping off", "why did checkout conversion fall this week", "build a signup-to-activation funnel", or "is mobile converting worse than desktop". Do NOT use for rewriting page copy or layout to lift a known weak step - use landing-page-cro instead; for decomposing active-user growth into new, retained, resurrected, and churned - use growth-accounting instead; for choosing which events to instrument and which product metrics to track - use product-analytics instead; for designing and reading a controlled experiment - use ab-test-analyzer instead.
What it loads into context
Frontmatter
170 tokensAlways available to the agent
SKILL.md
1,819 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: filled funnel with diagnosis
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,075 bytesSKILL.mdcore8,075 bytesSKILL.mdcore8,075 bytes
Read SKILL.md
# Funnel Analysis
A funnel tells you where money leaks out of a multi-step flow, but only if the definition is precise and the differences you report are statistically real. The costly mistake this skill prevents is shipping a "mobile checkout is broken" diagnosis built on an ambiguous funnel definition and a 40-user sample - teams reorganize roadmaps around noise. The output is a filled funnel table, one named bottleneck step, and the evidence that the gap is real.
## Operating procedure
Follow the steps in order: the funnel definition (Step 1) determines every count downstream, and significance testing (Step 5) must come before any diagnosis leaves the room.
### Step 1: Gather inputs and pin the definition
Collect these before touching SQL. If the user cannot answer, apply the default and label it a guess.
1. Ordered steps, as event names (e.g. view -> add_to_cart -> checkout -> purchase). Write them down verbatim; ambiguous definitions produce misleading funnels.
2. Conversion window: how long after step 1 a user has to finish. Default: the 90th percentile of observed time-to-convert, rounded up to a clean unit. Typical defaults if no data yet: 1 day for e-commerce checkout, 7 days for signup-to-activation, 14-30 days for B2B trial-to-paid.geo-places
When your agent uses it
Use when a task needs live geographic lookups - "geocode this address", "what's at these coordinates" (reverse geocoding), "lat/lon for this city", "which country/state is this ZIP or postal code in", or "country facts: capital, currency, population, flag". Nominatim (OpenStreetMap) is the geocoding default; Zippopotam for postal codes; APICountries for country facts. All keyless. Do NOT use for weather at a location - use weather-climate instead; do NOT use for country-level statistics over time (GDP, population trends) - use government-open-data instead; if the request is a vague "I need live data", route through public-data-api-picker.
What it loads into context
Frontmatter
173 tokensAlways available to the agent
SKILL.md
1,504 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Ranked APIs
- Procedure
- URL templates + real response shapes
- Deliverable
- Do NOT use - reach for X instead
- Quality bar
Files
SKILL.mdcore5,895 bytes
Read SKILL.md
# Geo & Places Data
Turn place names, addresses, coordinates, and postal codes into structured geography with keyless APIs. The mistake this prevents: reaching for the Google Maps/Places API from training memory - key-walled, billing-enabled - when Nominatim answers the same lookup free, or stalling on which of the dozens of directory-listed geocoders actually works.
## Ranked APIs
1. **Nominatim** (nominatim.openstreetmap.org) - DEFAULT for geocoding and reverse geocoding. OSM data, global, no key. Hard etiquette rules below.
2. **Open-Meteo Geocoder** - city/town name → lat/lon only. Faster and rule-free; use it when the input is a simple settlement name rather than an address or POI.
3. **Zippopotam.us** - postal code → place/state/coordinates, 60+ countries, no key.
4. **APICountries** (www.apicountries.com) - country facts: capital, currencies, languages, population, flags, borders. No key. (It serves the old REST Countries response shape; REST Countries itself now returns deprecation errors keyless - see Do NOT.)
## Procedure
1. **Classify the input**: address/POI → Nominatim search; coordinates → Nominatim reverse; bare city name → Open-Meteo geocoder; postal code → Zippopotam; country name/code → APICountries.ghostwriter
When your agent uses it
Analyzes an author's writing samples, builds an explicit quantified voice profile, and produces new content indistinguishable from the author's own hand. Use when someone asks "write this in my voice", "ghostwrite my newsletter", "make this sound like me", "draft the next chapter the way I would write it", or supplies writing samples plus a brief for a new piece. Do NOT use for drafting a persuasive opinion piece to pitch to a publication - use op-ed-writer instead. Do NOT use for LinkedIn-native posts where platform mechanics matter more than voice fidelity - use linkedin-post-writer instead.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,644 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Voice profile template
- Register contrast example
- Deliverable
- Do NOT
- Quality bar
- Escalation and adjacent jobs
Files
SKILL.mdcore7,389 bytes
Read SKILL.md
# Ghostwriter
Write in someone else's voice so completely that the reader never senses a second hand. Success is invisibility. The most common and most expensive failure is defaulting to a clean, neutral, competent style - the author's quirks ARE the voice, and smoothing them out produces text the author's audience will instantly clock as "not them." Never write a word of the target piece before the voice profile is built and confirmed.
## Operating procedure
Follow the steps in order. Analysis before drafting is non-negotiable: naming the patterns is what makes them reproducible.
### Step 1: Gather inputs
Collect from the user:
1. Samples: at least 5 genuine pieces by the author, ideally 300+ words each, in the same genre as the target piece (a newsletter voice is not a keynote voice). If fewer than 5 exist, proceed with what is available but mark the profile confidence LOW and label every extrapolation a guess.
2. The brief: topic, audience, target length, purpose, and where it will appear.git-commit-writer
When your agent uses it
Write commit messages from a staged diff in Conventional Commits 1.0.0 format - correct type and scope, `!` and BREAKING CHANGE footers, GitHub closing keywords, and a body that explains why - and split mixed changes into separately revertable commits. Use when someone asks "write a commit message for this diff", "commit this", "is this a breaking change?", or "split this into proper commits". Do NOT use for the pull request description - use pr-description-writer instead - or for turning a release's commits into a CHANGELOG.md - use changelog-generator instead.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
2,158 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Thresholds
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,981 bytes
Read SKILL.md
# Git Commit Writer
A commit message is the only documentation guaranteed to travel with the code: `git blame` surfaces it at the exact line someone is about to change, years after the ticket tracker is gone. The costly mistake this skill prevents is the message that restates the diff ("update user service", "fix bug", "wip") and drops the one thing the diff cannot show - why. A close second is the mixed commit, where a refactor, a fix, and a formatting sweep share one SHA and none can be reverted or bisected alone.
## Operating procedure
### Step 1: Read the actual change, not the file list
Run `git diff --staged` (or `git show <sha>` when rewording) and read every hunk. Collect: what behavior changed, which component owns it, whether a public surface changed (exported API, CLI flag, config key, HTTP contract, schema, event payload), and the motivation. Motivation is rarely in the diff - get it from the user, the issue, or the branch name, and flag it if inferred.
Read `git log --oneline -20` for conventions: Conventional Commits or not, existing scopes, trailers like `Signed-off-by:` (DCO). Match the repo; without Conventional Commits, write a plain imperative subject with the same body discipline.
### Step 2: Decide whether this is one commit or several
github-actions
When your agent uses it
Designs and hardens GitHub Actions CI/CD - pipelines that hit a sub-10-minute PR feedback budget through caching and parallelism, with least-privilege security and safe deploy gates. Use when someone asks "why is my CI so slow", "how do I cache node_modules properly", "set up a deploy workflow", "is pull_request_target safe here", or wants a workflow written or reviewed. Do NOT use for provisioning the cloud infrastructure the pipeline deploys to - use terraform-expert instead; do NOT use for Vercel-specific deploy flows - use vercel-deploy-pipeline instead; for secret storage policy beyond CI, use secrets-hygiene.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,779 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: workflow skeleton
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,481 bytes
Read SKILL.md
# GitHub Actions
CI that takes 25 minutes gets ignored, and a workflow with a write-token running untrusted code gets owned. This skill builds pipelines against two hard constraints at once - a speed budget developers will actually wait for, and a security posture that assumes PR authors are hostile - because retrofitting either into a sprawling workflow file is far more painful than starting right.
## Operating procedure
### Step 1: Gather inputs
- Stack and package manager (locks the caching strategy).
- Current wall-clock time of the slowest required check, and the test-suite runtime.
- Monorepo or single package; deploy targets and who may approve prod.
- Any use of forked PRs or `pull_request_target` today (security review trigger).
### Step 2: Set the budgetsgo-to-market-planner
When your agent uses it
Build a sequenced go-to-market plan - beachhead ICP, positioning, one or two channels, and dated milestones - instead of a scattershot launch. Use when someone asks "how do I take this to market", "what's my GTM strategy", "which channel should I start with", "who should I sell to first", or is planning a new product or a new market entry. Do NOT use for the dated launch-week timeline and asset checklist - use launch-plan-sequencer instead; for Product Hunt copy and day-of tactics use product-hunt-launch; for deep persona research use icp-persona-builder.
What it loads into context
Frontmatter
148 tokensAlways available to the agent
SKILL.md
1,514 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,706 bytes
Read SKILL.md
# Go-To-Market Planner
A GTM plan is a series of explicit bets about who you serve, why they switch, and where you reach them. The costly mistake this skill prevents is the scattershot: five channels, three audiences, no milestone that would prove any bet wrong - which burns six months of runway learning nothing. Sequence beats simultaneity: concentrate force on one beachhead, win it, then expand.
## Operating procedure
Follow the order - each step inherits the previous one. A channel chosen before the ICP is a guess; a milestone set before the channel is a wish.
### Step 1: Gather inputs
Collect before planning; use the defaults and label guesses as guesses.
1. The product and the painful moment it removes (one sentence).
2. Any existing traction: users, revenue, waitlist, warm audience. Default: zero.goals-accountability
When your agent uses it
Turns vague intentions into SMART goals with both lagging targets and controllable leading indicators, then keeps them alive with a 10-minute weekly check-in template, an accountability structure (visible goals, public commitments, partner check-ins), and a quarterly reflection that converts the quarter into lessons. Use when someone says "help me set goals I'll actually stick to", "I set resolutions and abandon them by February", "build me a goal tracking system", or wants a weekly review cadence for personal targets. Do NOT use for company or team objective-setting - use okr-builder instead - or for building daily habits - use habit-builder instead.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
794 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Step 1: Write SMART goals
- Step 2: Separate lagging from leading indicators
- Step 3: Weekly check-in (10 minutes)
- Step 4: Accountability structure
- Step 5: Quarterly reflection
- Tips
- Anti-patterns
Files
SKILL.mdcore3,432 bytes
Read SKILL.md
# Goals & Accountability
Goals without a tracking system are wishes. This skill turns intentions into SMART
goals, then keeps them alive with weekly check-ins and quarterly reflection.
## Step 1: Write SMART goals
Every goal must be:
- **Specific** - names exactly what will change.
- **Measurable** - has a number you can verify (the target).
- **Achievable** - a stretch, but realistic with effort.
- **Relevant** - tied to a larger objective you care about.
- **Time-bound** - has a deadline.google-business-profile-optimizer
When your agent uses it
Optimizes a Google Business Profile as the storefront of a local service business - primary category selection, service-area setup, weekly photo cadence, Q&A seeding, posts, and NAP consistency - ordered by ranking impact. Use when an owner asks "why don't I show up on Google Maps", "how do I set up my Google Business Profile", "what category should my business be", or "how do I get more calls from Google". Do NOT use for broader local ranking strategy across the website and citations - use local-seo-playbook instead. Do NOT use for review generation and responses - use review-engine instead.
What it loads into context
Frontmatter
135 tokensAlways available to the agent
SKILL.md
1,882 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to elicit
- Concrete thresholds
- Worked artifact: Summit Plumbing profile settings
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore8,735 bytes
Read SKILL.md
# Google Business Profile Optimizer
For a local service business, the Google Business Profile is the storefront:
most customers decide whether to call from the map pack panel without ever
visiting the website. The costly mistake this skill prevents is treating GBP
as a set-and-forget listing - a wrong primary category or an inconsistent
business name silently caps every dollar spent elsewhere. Before investing
hours here, confirm with local-unit-economics what a booked call is worth;
for Summit Plumbing (4 techs, $450 average ticket, $75,000 a month), each
extra customer from the profile is worth about $450 in first-job cash against
a roughly $119 paid CAC, so free GBP calls are the cheapest customers the
business will ever get.
## Operating proceduregovernment-open-data
When your agent uses it
Use when a task needs official country-level statistics - "GDP / population / life expectancy / inflation / unemployment for country X over time", "compare indicator Y across countries", or EU-official figures ("Eurostat says…"). World Bank Indicators is the default (no key, every country, 1,400+ series back decades); Eurostat covers EU-official statistics. Do NOT use for a country's static facts like capital or currency - use geo-places instead; do NOT use for live FX rates or crypto - use finance-fx instead; if the request is a vague "I need live data", route through public-data-api-picker.
What it loads into context
Frontmatter
148 tokensAlways available to the agent
SKILL.md
1,628 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Ranked APIs
- Procedure
- URL templates + real response shapes
- Deliverable
- Do NOT use - reach for X instead
- Quality bar
Files
SKILL.mdcore6,189 bytesSKILL.mdcore6,189 bytes
Read SKILL.md
# Government & Open Data
Pull official statistics from the World Bank and Eurostat keyless. The mistakes this prevents: quoting a GDP figure from training memory when the real series is one call away, and the two structural traps - World Bank's off-by-default JSON and its indicator-code lookup step - that make first attempts fail.
## Ranked APIs
1. **World Bank Indicators** (api.worldbank.org/v2) - DEFAULT. ~1,400 indicators × every country × back to 1960. No key. The catch: you query by opaque indicator code (`NY.GDP.MKTP.CD`), so the lookup step below is mandatory for unfamiliar metrics.
2. **Eurostat** (ec.europa.eu/eurostat/api) - EU-official statistics when the user wants the EU source of record or EU-only granularity. No key, JSON-stat format (denser to parse - shape below).
## Procedure
1. **Resolve the indicator code first.** Verified ones to use directly: `NY.GDP.MKTP.CD` (GDP, current US$), `SP.POP.TOTL` (population), `SP.DYN.LE00.IN` (life expectancy), `FP.CPI.TOTL.ZG` (inflation %), `SL.UEM.TOTL.ZS` (unemployment %). Anything else: search `https://api.worldbank.org/v2/indicator?format=json&per_page=50&source=2` filtered client-side, or hit `/v2/indicator/{guess}?format=json` to confirm a code exists before building the real query. Never invent a code - a bad code returns an error object inside a 200. Emissions codes (`EN.ATM.CO2E.PC`, `EN.GHG.*`) are served from a chronically failing backend - expect 502s/hangs and fall back to Eurostat (`env_air_gge`) for EU countries rather than retry-looping.
2. **Always append `format=json`.** The World Bank default is XML; forgetting this is the #1 failure.
3. **Prefer `mrv=N` over `date=` for "latest" queries.** `mrv=10` (most recent 10 values) hits the API's cache and skips the null-trimming problem; `date=` ranges on less-trafficked indicators frequently hang or 502 where the same indicator's `mrv` query answers instantly.grand-slam-offer-builder
When your agent uses it
Build or sharpen a gym's core offer into a one-page canvas - Value-Equation score, problem-to-solution stack, scarcity/urgency/bonus amplifiers, and a MAGIC name - so the right prospect feels stupid saying no. Use when an owner says "build my offer", "my offer isn't converting", "what should I include", "make my offer irresistible", "design my challenge offer", or "name my program". Do NOT use when the task is choosing the price point or writing guarantee language - use gym-pricing-and-guarantees instead; do NOT use when the task is the day-by-day run-of-show for a challenge - use gym-transformation-challenge instead.
What it loads into context
Frontmatter
151 tokensAlways available to the agent
SKILL.md
2,653 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: offer-canvas
- references/value-equation
- references/offer-construction
- references/naming-magic
Files
SKILL.mdcore11,222 bytesSKILL.mdcore11,222 bytesSKILL.mdcore11,222 bytes
Read SKILL.md
# Grand Slam Offer Builder
Turn a gym's front-end challenge and core membership into offers so strong the right prospect feels stupid saying no, by raising delivered value and lowering the cost in time and effort, then stacking solutions until perceived value dwarfs the price. Deliverable: a filled one-page offer canvas.
## Workflow
Run these steps in order. Do not skip the Value-Equation score; it tells you what to fix first.
1. **Score the current offer on the Value Equation.** Value rises with the dream outcome and the prospect's belief they will reach it, and falls with the time and effort it costs: `Value = (Dream Outcome x Perceived Likelihood) / (Time Delay x Effort and Sacrifice)`. Rate the offer 1-10 (10 = best) on four levers: Dream outcome (how badly they want the result), Perceived likelihood (how believable the result is for them), Speed (how fast results feel - inverse of time delay), Ease (how little effort it demands - inverse of effort). Run `value_score.js` (below). It multiplies the four and names the lowest lever. Fix that lever first: the equation is multiplicative, so the weakest lever caps the whole offer no matter how strong the others are. See references/value-equation for how to raise each one.
2. **Build the offer from problems, not features.** Name the dream outcome in the prospect's own words ("drop two dress sizes before the wedding and keep it off"). List every problem on the path (no time, no idea what to eat, intimidated by the gym floor, tried before and quit, no accountability). Turn each problem into a solution stated as the thing you provide ("no idea what to eat" → "a done-for-you meal plan with a swap list"). Choose a delivery vehicle for each (group session, 1:1 check-in, app, printed guide, private community, text support) that fits the value and your cost to deliver. Trim and stack: cut anything low-value or expensive, then stack the survivors into one offer that reads as overwhelming value for the price. See references/offer-construction for full worked examples on both the challenge and the membership.
3. **Enhance with the four amplifiers, only after the core stack is strong.** They multiply a good offer and do nothing for a weak one. Scarcity: a real capacity limit ("20 challengers per cohort because that is what our coaches can serve well"). Urgency: a real deadline ("cohort starts the 6th; doors close the 3rd"). Bonuses: extra wins that each remove a specific objection, each with a name and a stated value. Guarantee: hand off to gym-pricing-and-guarantees to choose the type and write the language, then drop the result into the canvas guarantee slot.
grant-loi-writer
When your agent uses it
Writes the one-page letter of inquiry that earns a nonprofit an invitation to submit a full proposal, using the need-approach-fit-ask structure, funder-language mirroring, and a specific ask stating amount, duration, and what the money buys. Use when a nonprofit leader says "write a letter of inquiry", "this funder wants an LOI first", "how do I introduce our program to a foundation", or "turn our program summary into a one-pager for a funder". Do NOT use for the full grant proposal - use grant-proposal-writer instead - and do NOT use for individual donor appeal letters - use donor-communications instead.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
1,880 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: Bright Futures LOI skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,378 bytesSKILL.mdcore8,378 bytes
Read SKILL.md
# Grant LOI Writer
A letter of inquiry has one job: earn the invitation to submit a full proposal. It is not a compressed proposal, and the most common failure is treating it as one - cramming in the logic model, the evaluation plan, and the budget until the letter runs three pages and the program officer stops reading. This skill produces a one-page LOI that a program officer can champion internally in ninety seconds. It sits second in the grants pipeline: it takes a pursue-tier prospect from grant-prospect-researcher and, on invitation, hands off to grant-proposal-writer.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students, asking the Hearthstone Foundation for $75,000.
## Operating procedure
### Step 1: Confirm the prospect is qualified
Do not write an LOI for a funder that has not passed fit-scoring in grant-prospect-researcher (score 11+ of 15). Writing to unqualified funders is how three-person teams burn their application capacity. Pull from the prospect table: the funder's median grant, its stated priorities, two or three named similar grantees, and any submission requirements (word limits, portal, required attachments).
### Step 2: Harvest funder language for mirroring
grant-proposal-writer
When your agent uses it
Writes the full foundation grant proposal for a nonprofit - need statement grounded in local data, a logic model with real numbers (inputs to activities to outputs to outcomes), SMART objectives, an evaluation plan sized to organizational capacity, a budget narrative where every line traces to an activity, and a credible sustainability answer. Use when a nonprofit leader says "write the full grant proposal", "we got invited to submit", "build our logic model", "our proposal needs an evaluation plan", or "the funder wants a budget narrative". Do NOT use for the initial one-page letter of inquiry - use grant-loi-writer instead - and do NOT use for corporate sales proposals - use sales-proposal-writer instead.
What it loads into context
Frontmatter
151 tokensAlways available to the agent
SKILL.md
2,128 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: Bright Futures logic model
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore10,408 bytes
Read SKILL.md
# Grant Proposal Writer
The full proposal is where invited organizations lose grants they should win: national statistics standing in for local need, outcomes that are really activities, an evaluation plan the organization cannot execute, and a budget that does not match the narrative. Program officers read for exactly these four failures. This skill builds the proposal in dependency order so each section is forced to agree with the others. It is third in the grants pipeline: it starts from the LOI that grant-loi-writer got invited, and its numbers must match that LOI exactly - a proposal that contradicts its own LOI reads as an organization that does not know its numbers.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students, invited by the Hearthstone Foundation to propose for $75,000 to expand to 325 students.
## Operating procedure
Build in this order - logic model before objectives, objectives before evaluation, activities before budget - because each section is derived from the previous one, and writing them independently is how contradictions get in.
### Step 1: Fix the baseline numbers
Copy the LOI's numbers as immutable: ask amount, duration, what it buys, scale, headline outcome. Collect the rest: program costs by line, staff roles and salaries allocated to the program, outcome data from the last two years with sources, and the local need data. Label every guessed number as a guess and resolve it before submission.
grant-prospect-researcher
When your agent uses it
Finds and qualifies foundation funders for a nonprofit by fit-scoring prospects on mission alignment, typical grant size, geographic focus, and past-grantee similarity, reading the funder's 990 for real priorities and median grant size, and sizing the pipeline at 4-5x the annual grant goal. Use when a nonprofit leader says "find foundations that would fund us", "is this funder worth applying to", "how many grants do we need in the pipeline", or "what does this funder's 990 tell us". Do NOT use for B2B sales prospect lists - use prospect-list-builder instead - and do NOT use for startup investor research - use investor-targeting instead.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,968 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: Bright Futures prospect-scoring table
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,643 bytes
Read SKILL.md
# Grant Prospect Researcher
Small nonprofits lose most grants before a word of the proposal is written: they apply to funders who were never going to fund an organization of their size, in their geography, doing their work. Every poorly fit application burns two to four staff-weeks that a three-person team does not have. This skill qualifies funders before any writing starts, so the writing time goes only to prospects with a real 20-30 percent chance of winning. It is the first skill in the grants pipeline: qualified prospects flow to grant-loi-writer, and invitations from there flow to grant-proposal-writer.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students, seeking a $75,000 foundation grant toward a $180,000 annual grant-revenue goal.
## Operating procedure
Follow the steps in order. Pipeline sizing comes first because it tells you how many qualified prospects you need before you know whether the list is done.
### Step 1: Set the grant goal and size the pipeline
Pipeline sizing rests on one practitioner number, defined here and cited by nonprofit-board-pack: well-fit prospects win at roughly 20-30 percent, so the pipeline of submitted asks must total 4-5x the grant goal.
graphql-schema
When your agent uses it
Designs GraphQL schemas and resolvers that scale - domain-modeled types, Relay pagination, DataLoader batching to kill N+1, mutation payloads with typed user errors, and depth/complexity limits that stop abusive queries. Use when someone asks "how should I structure this GraphQL type", "my resolvers are hammering the database", "cursor or offset pagination", "how do I version a GraphQL API", or is designing or reviewing a schema or federation split. Do NOT use for REST or RPC endpoint design - use api-design instead; do NOT use for the underlying table design - use database-schema instead; for hunting existing N+1s in a codebase, use n-plus-one-hunter.
What it loads into context
Frontmatter
162 tokensAlways available to the agent
SKILL.md
1,796 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: schema design, bad vs good
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,017 bytesSKILL.mdcore8,017 bytes
Read SKILL.md
# GraphQL Schema
A GraphQL schema is a public contract that clients build against for years, and a resolver layer is an open invitation for N+1 queries and abusive nesting. This skill designs schemas clients love and servers survive: model the domain rather than the database, batch every relationship resolver, and put hard limits on what a single query may cost - because removing a shipped field or fixing a hot N+1 under load is 10x the effort of designing it right.
## Operating procedure
### Step 1: Gather inputs
- The domain nouns and the operations clients actually need (screens/use-cases, not tables).
- Largest list sizes and expected query shapes; who the clients are (first-party only vs public API - public means stricter limits).
- Data sources behind each type and which relationships cross a database or service boundary.
- Single graph or federated subgraphs.
### Step 2: Model the graphgrowth-accounting
When your agent uses it
Decomposes active-user or revenue growth into new, retained, resurrected, expansion, contraction, and churned flows, reconciles the identity exactly, and reads the quick ratio to diagnose whether growth is an acquisition, retention, or resurrection problem. Use when someone asks "why did our MAU change", "our top line grew but I don't trust it", "what's our quick ratio", "build an MRR bridge", or "is our growth acquisition-driven or retention-driven". Do NOT use for forecasting future growth from loops and assumptions - use growth-model instead - for general event instrumentation and funnel metrics - use product-analytics instead - or for churn interventions themselves - use churn-reduction instead.
What it loads into context
Frontmatter
158 tokensAlways available to the agent
SKILL.md
1,740 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: growth-accounting table
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,346 bytes
Read SKILL.md
# Growth Accounting
A top-line active-user or revenue chart can rise while the business underneath rots: heavy acquisition masking accelerating churn is the classic failure, and it is invisible until the paid channel stalls. Growth accounting decomposes every period's change into its flows so the chart cannot lie, and the decomposition tells you which lever - acquisition, retention, or resurrection - actually needs work.
## Operating procedure
Reconcile before you interpret: an identity that does not balance means the flags are wrong, and every downstream conclusion with them.
### Step 1: Gather inputs
1. The activity definition - what user action counts as "active." Require a meaningful action, not just a login; default to the product's core value event.
2. The period - daily, weekly, or monthly. Match it to the natural usage cadence and keep it consistent; a monthly product measured weekly will show fake churn.
3. The event table - one row per user per active period (or MRR per account per month for revenue).
4. Whether the user wants user-count accounting, revenue accounting (the MRR bridge), or both. Default: both if MRR exists.growth-model
When your agent uses it
Builds a driver-based growth model - acquisition, activation, retention, monetization, and a loop factor - that projects users and revenue bottom-up, runs base/upside/downside scenarios, and names the one constraint to work on next. Use when someone asks "build me a growth model", "what should our user projection be", "which lever moves growth most", "model our viral loop", or wants projections driven by real inputs instead of a hockey-stick guess. Do NOT use for the SaaS MRR bridge, NRR, and revenue-forecasting mechanics - use revenue-modeling. Do NOT use for decomposing historical active-user change into new, retained, resurrected, and churned - use growth-accounting. Do NOT use for per-customer CAC/LTV math - use unit-economics.
What it loads into context
Frontmatter
180 tokensAlways available to the agent
SKILL.md
2,186 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Calculator
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore9,174 bytesSKILL.mdcore9,174 bytesSKILL.mdcore9,174 bytes
Read SKILL.md
# Growth Model
A growth model is not a revenue forecast with a hopeful curve. It is a driver-based system that shows how inputs (traffic, conversion, retention) compound into output (users, revenue) - and where the leverage is. The costly mistake this prevents: drawing the output curve first and back-filling assumptions to justify it, which produces a plan nobody can steer because no team owns an input.
## Operating procedure
Order matters: the loop must be identified before the equation is written, and the equation before any projection - otherwise the model quietly becomes a fixed-percent guess with extra columns.
### Step 1: gather inputs
- Current active users and the definition of "active" (a meaningful action, not a login).
- New users per period by channel, with each channel's volume and cost.
- Activation rate: the share of new users reaching first value. From real funnel data where it exists.
- Retention: the share of actives who stay each period - from cohort curves, never a flat assumption. If only aggregate churn exists, label the retention figure a guess.gtd-system
When your agent uses it
Runs the full Getting Things Done loop - capture, clarify, organize, reflect, engage - building a trusted system of context lists, a projects list with defined next actions, and a weekly review habit. Use when someone says "I'm overwhelmed and things are slipping through the cracks", "set up GTD for me", "help me do a brain dump and organize it", or "my to-do list is a mess". Do NOT use for just running the weekly review ritual itself - use weekly-review instead - or for clearing an email backlog - use inbox-zero.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
1,652 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: filled weekly review checklist
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,181 bytes
Read SKILL.md
# Getting Things Done
The goal is a clear head: every commitment captured in a system more trustworthy than memory, so the mind stops rehearsing open loops at 2 a.m. The costly failure this skill prevents is the half-adopted system - lists that exist but aren't trusted, so the user runs both the lists and the mental inventory, and gets the overhead of both with the relief of neither. Trust comes from one thing: the weekly review. A GTD setup without it decays into another abandoned app within 3 weeks.
## Inputs to collect
1. Current tools: where tasks, calendar, and notes live today. Default: keep the user's existing tools - GTD is tool-agnostic, and a tool migration on day one doubles the failure risk.
2. Open commitments: work, personal, waiting-on-others. Gathered in Step 1, not asked up front.
3. Contexts that fit their life (@calls, @computer, @errands, @home, @agenda-boss). Default: those five; label the guess.
4. A weekly review slot: a recurring 60-90 minute block. Non-negotiable - get a day and time on the calendar before finishing setup.
## Operating procedure
The five steps run in order because each consumes the output of the last: clarify processes what capture collected; organize files what clarify decided.gym-meta-ads-funnel
When your agent uses it
Write gym Meta (Facebook/Instagram) ad creative, structure the lead-to-consult funnel, set the budget, and read funnel costs against the challenge economics. Use when a gym owner asks to write Facebook/Instagram ads, says ads aren't converting or cost per lead is too high, asks what to spend on ads, or wants to build or scale a Meta ad funnel that fills a challenge or front-end offer. Do NOT use for general A/B creative-testing methodology - use ad-creative-testing instead; do NOT design the offer being sold or the consult close - use gym-transformation-challenge and closer-sales-script instead.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
3,074 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: ad-script-pack
- Template: landing-page-copy
- references/ad-anatomy
- references/funnel-and-tracking
- references/budget-and-scaling
Files
SKILL.mdcore12,180 bytesSKILL.mdcore12,180 bytesSKILL.mdcore12,180 bytes
Read SKILL.md
# Gym Meta Ads Funnel
Sell the challenge or front-end offer in the ad, route strangers into a short funnel that books a consult, and price the spend against what the offer can afford. The ad sells the challenge, never the gym itself.
## Workflow
1. **Set the offer and price first.** Ads amplify a good offer and cannot rescue a weak one. If the challenge, price, and guarantee are not defined, stop and define them (see gym-transformation-challenge for the offer, gym-pricing-and-guarantees for price and guarantee). Do not write a single ad before this is fixed.
2. **Build the shortest funnel that converts.** Every extra step loses people.
- Ad: sells the challenge with a clear hook and offer.
- Lead capture: a native Meta lead form (fast, low CPL, lower intent) or a landing page (higher intent, needs copy). Define the conversion event as the lead submission.
- Automated booking: the instant a lead submits, fire a text plus a booking link for the intro consult.
- Consult: the low-pressure intro where the challenge or membership is sold.
- Track the full path - ad to lead to booked to shown to sold - not just to lead.gym-money-model
When your agent uses it
Model gym unit economics - Client-Financed Acquisition, 30-day cash, CAC, LTV, LTV:CAC, and payback - and return a verdict on whether growth self-funds. Use when a gym owner asks "what's my CAC", "how much can I spend to get a member", "what should my challenge cost to break even", "is my gym profitable to scale", "what's my LTV:CAC", or "why does growth stall when ads work". Do NOT use for generic SaaS or non-gym business unit economics - use unit-economics instead.
What it loads into context
Frontmatter
131 tokensAlways available to the agent
SKILL.md
3,242 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Quality bar
- Calculator
- Template: unit-economics-sheet
- Do NOT
- references/money-model
- references/the-2x-cash-rule
- references/diagnosing-economics
Files
SKILL.mdcore13,068 bytesSKILL.mdcore13,068 bytesSKILL.mdcore13,068 bytes
Read SKILL.md
# Gym Money Model
Build this first. Every other gym growth skill spends money or sets a price, and
this model tells you whether that spend pays for itself. The goal is
Client-Financed Acquisition: the front-end collects enough cash in the first 30
days that each new member funds the acquisition of the next one, so growth is not
capped by the bank balance.
Work the example gym throughout: one location, semi-private coaching plus
memberships, a 6-week transformation challenge as the front-end.
## Operating procedure
Follow these steps in order. Do not skip to LTV before the front-end math isgym-pricing-and-guarantees
When your agent uses it
Set a single-location gym's price points and design a risk-reversal guarantee it can actually afford, with margin math so a refund never bleeds the business. Use when a gym owner asks "what should I charge", wants to set or raise a membership or challenge price, asks "is a money-back guarantee safe" or "how risky is my guarantee", or needs guarantee language to put on an offer. Do NOT use when constructing the whole offer (value stack, bonuses, scarcity, naming) - use grand-slam-offer-builder instead; do NOT use when testing whether the price self-funds acquisition - use gym-money-model instead.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
2,422 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: price-architecture
- Template: guarantee-language
- references/pricing-on-value
- references/guarantee-types
Files
SKILL.mdcore10,681 bytesSKILL.mdcore10,681 bytesSKILL.mdcore10,681 bytes
Read SKILL.md
# Gym Pricing and Guarantees
Price on the value and outcome you deliver - not on your delivery cost or the gym down the road - then bolt on a guarantee that removes the buyer's risk without putting the business at risk. This skill owns the gym's price points and guarantee language only.
## Workflow
### Step 1: Price on value, not cost or competitor
- Anchor high, then justify with the stacked value: a $499 challenge shown next to a list of inclusions worth far more reads as a deal; the same $499 with no framing reads as expensive.
- Never cost-plus. Adding a margin to delivery cost caps the price at your imagination of cost, not the buyer's desire for the outcome.
- Never match the local discount gym. Matching a budget competitor tells the buyer you are the same thing for less and destroys the premium position. A higher price can raise perceived likelihood - people trust that real results cost something.
- If you must move on price, add value or remove an inclusion - never cut the number. See references/pricing-on-value.
### Step 2: Set the three price points
A single-location gym usually runs three tiers. Set each with a target gross margin and a one-line value justification, then fill the price-architecture template with the gym's actual numbers.gym-transformation-challenge
When your agent uses it
Designs a gym's 6-week transformation challenge as a self-funding front-end offer and the week-by-week ascension that converts challengers into recurring members. Use when a gym owner asks to "design a 6-week challenge", structure a "transformation challenge", build a "front-end offer to fill my gym", plan a challenge "run-of-show", or figure out "how do I convert challengers to members". Do NOT use when the request is about the paid ads, audiences, or lead forms that fill the challenge - use gym-meta-ads-funnel instead.
What it loads into context
Frontmatter
127 tokensAlways available to the agent
SKILL.md
2,690 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: challenge-offer-sheet
- Template: conversion-offer-sheet
- Template: challenge-runbook
- references/challenge-blueprint
- references/challenge-economics
Files
SKILL.mdcore11,336 bytesSKILL.mdcore11,336 bytesSKILL.mdcore11,336 bytes
Read SKILL.md
# Gym Transformation Challenge
Structure a time-bound (usually 6-week) front-end challenge that fills the gym at low friction, breaks even on its own ad spend, and is engineered to ascend challengers into recurring membership.
## Workflow
### Step 1: Structure the challenge
Decide and write down each element:
- **Duration**: 6 weeks by default - long enough for a visible result, short enough to feel doable.
- **Promise**: one specific, believable result. "Drop a dress size in 6 weeks" beats "get fit."
- **Inclusions**: sessions per week, a done-for-you nutrition plan, weekly accountability check-ins, a start and finish body-composition scan, a private group. Each inclusion removes one obstacle the buyer would otherwise face.
- **Price**: common range 199 to 599. The price must cover ad spend and fulfillment, leaving the back-end membership as the profit. (If you also run gym-money-model, set the price there so the challenge clears the cash rule; otherwise size it directly with Step 4.)habit-builder
When your agent uses it
Turns a vague intention into a one-page habit plan - a specific implementation intention, a habit-stack anchor sentence, cue-craving-routine-reward loop mapping, environment design, a never-miss-twice tracking method, and weekly review questions. Use when someone says "help me build a habit", "I want to exercise more but can't stick with it", "why do my habits keep failing", or wants to break a bad habit by removing its cues. Do NOT use for goal-setting with an accountability structure - use goals-accountability instead; for a full training program, use fitness-program.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
747 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 1. Define the target behavior
- 2. Map the habit loop
- 3. Write an implementation intention
- 4. Stack the habit
- 5. Design the environment
- 6. Track without perfectionism
- 7. Review weekly
- 8. Scale gradually
- Output for the user
Files
SKILL.mdcore3,271 bytes
Read SKILL.md
# Habit Builder
Note: This is general self-improvement education, not professional, medical, or psychological advice. Consult a qualified professional for clinical concerns.
Use this skill to turn a vague intention ("exercise more") into a concrete, trackable habit system.
## 1. Define the target behavior
Make it specific and small. Bad: "read more." Good: "read one page after I pour my morning coffee."
Ask the user:
- What is the smallest version of this habit that still counts? (the "two-minute rule")
- What identity does this habit reinforce? ("I am someone who reads daily")
## 2. Map the habit loop
Every habit has four parts. Fill each one in explicitly:hashtag-strategist
When your agent uses it
Builds a tiered hashtag and keyword set sized to the account's actual reach - mixing broad, niche, and branded tags per platform limits and screening out banned, spammy, or shadow-flagged tags. Use when someone asks "what hashtags should I use", "build me a hashtag set for this post", "audit my tag block", or is creating reusable tag sets per content pillar on Instagram, TikTok, LinkedIn, X, or YouTube. Do NOT use for writing the caption itself - use social-caption-writer instead. Do NOT use for clustering SEO search keywords by topic or intent - use keyword-cluster-builder instead.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,459 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: one post
- Deliverable
- Do NOT
- Quality bar
- Adjacent jobs
Files
SKILL.mdcore6,269 bytes
Read SKILL.md
# Hashtag Strategist
Produce a tiered, platform-sized hashtag set plus in-caption keywords matched to the account's actual reach, not to tag volume. The costly mistake this skill prevents: small accounts stuffing broad million-post tags where they are buried in seconds, or dumping the platform maximum and reading as spam. Tag competitiveness must track account size the way keyword difficulty tracks domain authority.
## Operating procedure
### Step 1: Gather inputs
Collect before building anything; where a number is unknown, use the stated default and label it a guess.
1. Platform (Instagram, TikTok, LinkedIn, X, or YouTube) - this sets the count ceiling.
2. Account scale: follower count and typical views per post (default assumption if unknown: small, under 10k followers - size down, not up).
3. The specific post or the content pillars the sets will serve.
4. Brand tag and any community tags already in use.help-documentation
When your agent uses it
Writes user-facing help-center articles - verb-first how-to guides, symptom-organized troubleshooting pages, and FAQs - with one task per article, exact UI labels, and a stated success result. Use when someone asks "write a help article for...", "document how users reset their password", "turn these support tickets into a troubleshooting page", or "our help center articles are confusing, rewrite this one". Do NOT use for internal team procedures and SOPs - use process-doc instead; for on-call operational runbooks, use runbook-writer; for internal knowledge-base articles aimed at support agents, use kb-article-writer.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,333 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Skeleton: how-to article
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore5,940 bytes
Read SKILL.md
# Help Documentation
The reader of a help article arrives stuck, impatient, and looking for one specific answer; the job is to get them unstuck fast. The costly failure this skill prevents is the feature tour - an article organized around what the product has instead of what the user is trying to do, which the stuck user scrolls, fails to match to their problem, and abandons for a support ticket.
## Operating procedure
### Step 1: Gather inputs
1. The task or problem the article addresses, in the user's words (required). One article = one task; if the request covers several tasks, split into linked articles now.
2. The exact UI labels, menu paths, and screens involved (required - ask for screenshots or the live product; guessing labels produces instructions that fail on contact).
3. Audience and prerequisites: what plan/role/permissions the reader needs.
4. Platform or version differences that change the steps (default: none noted).
5. The most common failure points, from support tickets if available.
hiring-a-players-and-sops
When your agent uses it
Build a gym's role-outcome hiring funnel - scorecards, work-sample screening, A-player pay logic, hire-order sequencing - and turn the owner's work into handoff-ready SOPs. Use when a gym owner says "hire a coach", "find A-players", "first hire for my gym", "build an interview process", "document my processes", "write an SOP", or "get out of the day-to-day". Do NOT use when the user needs a generic, inclusive, legally-careful job posting for any company - use job-description-writer instead.
What it loads into context
Frontmatter
129 tokensAlways available to the agent
SKILL.md
2,335 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: role-scorecard
- Template: sop-template
- references/hiring-funnel
- references/sop-method
Files
SKILL.mdcore9,595 bytesSKILL.mdcore9,595 bytesSKILL.mdcore9,595 bytes
Read SKILL.md
# Hiring A-Players and SOPs
A gym stops scaling when the owner is the bottleneck. Run a role-outcome hiring funnel that finds people who produce results without supervision, and a four-step method that turns the owner's knowledge into SOPs someone else can run.
## Workflow
Pick the entry point: if the owner is drowning in operations, start at Step 5 (SOPs). If growth is capped by the owner's own selling or coaching hours, start at Step 1 (hiring).
### Step 1: Sequence the hire before defining it
A scaling single-location gym hires in this order:
1. Sales setter or salesperson first - book and close so the owner stops being the only seller. Frees the most owner time and funds the next hire.
2. Coach next - deliver sessions as the member base grows, protecting retention.hiring-pipeline
When your agent uses it
Designs an end-to-end hiring pipeline from job requisition through debrief and offer, with stage-by-stage conversion benchmarks, a reverse-funnel capacity plan, and time-to-fill targets. Use when someone asks "design our hiring process", "how many candidates do we need to make one hire", "our offers keep getting declined", "candidates are dropping out of our funnel", or "how long should hiring take". Do NOT use for writing the interview questions and rubrics for one loop - use interview-guide-builder instead - for writing the sourcing messages to candidates - use candidate-outreach-personalizer instead - for the scorecard artifact itself - use hiring-scorecard instead - or for the job post copy - use job-description-writer instead.
What it loads into context
Frontmatter
166 tokensAlways available to the agent
SKILL.md
1,673 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: pipeline conversion table
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,695 bytesSKILL.mdcore7,695 bytesSKILL.mdcore7,695 bytes
Read SKILL.md
# Hiring Pipeline
An unstructured hiring process produces slow decisions, biased ones, and lost candidates - the top candidate accepts elsewhere during week three of a meandering loop. A structured pipeline reduces bias, compresses time-to-hire, and gives every candidate a consistent experience; the reverse funnel below also tells you, before you open the role, whether the top of the funnel is big enough to produce a hire at all.
## Operating procedure
The order is load-bearing: the scorecard (Step 1) must exist before sourcing, because every later stage evaluates against it, and interviewers cannot own attributes that were never defined.
### Step 1: Write the requisition and scorecard
Collect from the hiring manager before anything opens:
- Mission of the role - what success looks like in 6-12 months, one paragraph.
- Must-have vs nice-to-have skills - keep must-haves to 3-5. Long lists deter great candidates, especially from underrepresented groups.hiring-scorecard
When your agent uses it
Builds structured interview scorecards tied to role-specific signals to reduce bias and improve calibration. Use when opening a new role or auditing an existing interview loop for consistency.
What it loads into context
Frontmatter
51 tokensAlways available to the agent
SKILL.md
915 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Step 1 - Define the Role's Top 4-6 Signals
- Step 2 - Map Signals to Interview Slots
- Step 3 - Write the Scoring Rubric
- Step 4 - Require Evidence, Not Ratings
- Step 5 - Run a Calibrated Debrief
- Escape Hatches
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore4,373 bytes
Read SKILL.md
# Hiring Scorecard Builder
Unstructured interviews have poor predictive validity and amplify the biases of whoever talks most. A scorecard forces interviewers to decide what matters before meeting the candidate and to gather evidence, not impressions.
## Step 1 - Define the Role's Top 4-6 Signals
Start from the first 90-day outcomes, not a list of desired traits. Ask: what does success look like concretely at 30, 60, and 90 days? Then work backward to the signals a candidate would have to demonstrate to make those outcomes likely. Good signals are observable in a conversation. Bad signals are adjectives ('smart', 'passionate', 'culture fit').
## Step 2 - Map Signals to Interview Slots
Each interviewer owns 1-2 signals. Never have two interviewers independently assess the same signal with no calibration plan - duplication wastes time and creates conflicting data. Assign technical depth signals to engineers, cross-functional signals to PMs or peers, and leadership/judgment signals to the hiring manager.
## Step 3 - Write the Scoring Rubric
icon-system
When your agent uses it
Designs or audits an icon set so every glyph shares one grid, stroke weight, corner language, naming convention, and export pipeline, and delivers the written icon spec engineering builds against. Use when someone asks "our icons look inconsistent", "set up an icon grid", "how should we name icons", "audit this icon set", "prep our SVGs for handoff", or is commissioning new icons and needs the rules first. Do NOT use for a single logo mark - use logo-brief-writer instead; for delivering full design specs beyond icons, use design-handoff-doc.
What it loads into context
Frontmatter
133 tokensAlways available to the agent
SKILL.md
1,851 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked icon spec (copy and fill)
- Consistency audit checklist
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,817 bytes
Read SKILL.md
# Icon System
An icon set without shared rules decays one icon at a time: each new glyph is drawn against a slightly different grid, stroke, or metaphor, and within a year the product carries three visual dialects that no single redraw can fix cheaply. This skill produces the spec that makes every icon - including ones drawn by future contributors - feel like one family, and the audit checklist that catches drift before it ships.
## Operating procedure
Follow the order: the grid constrains the stroke, the stroke constrains the drawing rules, and naming/export only matter once the drawing rules are fixed. Changing the grid after 200 icons exist means redrawing 200 icons.
### Step 1: Gather inputs
Collect before drawing anything. Label assumptions as guesses.
1. Rendering sizes in product (commonly 16, 20, 24px). The smallest real size drives detail decisions.
2. Style direction: outline, filled, or both. If both, they are two coordinated sets, not one.icp-persona-builder
When your agent uses it
Turns a vague "who we sell to" into a weighted, filterable ICP scorecard - firmographics, technographics, and time-bound triggers expressed as literal Apollo or Sales Navigator filter values, built from closed-won and churned evidence, with hard disqualifiers and A/B/C account tiering - plus a buying-committee persona map covering economic buyer, champion, technical user, and blocker with each role's pains. Use when standing up outbound - FIRST, before any list-building or outreach - and someone asks "who is our ICP", "what filters do I put in Apollo", "we're targeting everyone and closing no one", or "score this account against our ICP". Do NOT use for pulling the actual prospect list - use prospect-list-builder instead - or for the messaging one-liner - use positioning-statement instead - or for end-user research personas - use user-persona instead.
What it loads into context
Frontmatter
191 tokensAlways available to the agent
SKILL.md
2,871 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The workflow
- ICP scorecard template
- Worked example - "Sentinel" (fictional incident-response SaaS)
- Runnable artifact - score an account against weighted ICP
- Common failure modes
Files
SKILL.mdcore11,608 bytesSKILL.mdcore11,608 bytesSKILL.mdcore11,608 bytes
Read SKILL.md
# Build Your ICP & Buyer Personas
The core insight: an ICP is not a description, it is a **filter**. If you cannot type a criterion into Apollo, Sales Navigator, or a Clay table and get a finite list back, it is not part of your ICP - it is marketing prose. "Mid-market companies that value innovation" is useless. "US-based B2B SaaS, 50-500 employees, Series B+, using Salesforce + Segment, hiring SDRs" is a list you can build tomorrow. The most common trap is conflating the ICP (the *account* you target) with the persona (the *human* you email) and skipping the buying committee entirely - so reps email one champion who has no budget, or one economic buyer who has never felt the pain.
The second trap is building the ICP from aspiration instead of evidence. Your ICP is whoever already gets value and pays you, not whoever you wish would. Start from closed-won, not from the TAM slide. If you have fewer than ~10 closed deals, anchor on your best 3-5 customers by retention and expansion, and treat the result as a hypothesis to be falsified by [[buying-signal-tracker]] reply data.
## When to use this skill
- You are standing up outbound and have no documented, filterable target definition.
- Reps are spraying - pipeline is full of bad-fit deals that stall at procurement or "no budget."
- Reply rates are fine but win rates are bad: usually a persona/committee problem, not a copy problem.
- You're entering a new segment or launching a new product and need a fresh ICP hypothesis.
- You need to tier a raw account list A/B/C before [[prospect-list-builder]] and [[outreach-sequence-designer]] burn effort on it.
idempotency-enforcer
When your agent uses it
Designs client-supplied idempotency keys, deduplication storage, and replay semantics so at-least-once delivery and client retries return the first result instead of double-charging or double-shipping. Use when someone asks "how do I stop duplicate charges on retry", "add an Idempotency-Key header to this endpoint", "the queue redelivered a job and we shipped twice", or before exposing any unsafe POST behind a retrying client or queue. Do NOT use for inbound third-party webhook handlers whose dedup is keyed on a provider event ID - use webhook-receiver-hardener instead and reference this skill only for the storage pattern; do NOT use for designing the client-side retry/backoff policy itself - use rate-limit-handler instead.
What it loads into context
Frontmatter
174 tokensAlways available to the agent
SKILL.md
1,787 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: middleware design
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,161 bytes
Read SKILL.md
# Idempotency Enforcer
Make the second arrival of any non-reversible write a no-op that replays the first result. Without this, every retrying client, queue redelivery, and impatient double-click is a potential duplicate charge, duplicate shipment, or duplicate email - and the failure only shows up in production, under load, as a customer complaint.
## Operating procedure
Steps 1-3 are design decisions; 4-7 are the runtime protocol. Do them in order - the storage decision (step 3) depends on the scope decision (step 2), and the replay rules (step 6) are meaningless until intent is persisted atomically (step 5).
### Step 1: Gather inputs
Collect before designing. Label guesses as guesses.
- Which endpoints have non-reversible side effects (payments, orders, emails, inventory)? Only those get the machinery.
- Is the side effect in the same database as the API (one transaction possible) or an external call (a charge, a third-party API)?
- What is the client's maximum retry window - how long after the first attempt can a retry legitimately arrive? Default assumption: 24 hours.image-license-rights
When your agent uses it
Decides whether a specific image use is legally permitted, what the license obligates, and whether a model or property release is required - covering CC0 and Creative Commons variants, Unsplash/Pexels/Pixabay terms, royalty-free vs rights-managed, standard vs extended tiers, editorial-use-only, and AI-image caveats - and delivers a per-image license record. Use when someone asks "is this free for commercial use", "do I need to credit the photographer", "can I put this on merchandise", "what is royalty-free vs rights-managed", "is editorial-use-only a problem", or "do I need a model release", or is about to publish, advertise, or sell anything containing a stock image. Do NOT use to find or search for a photo - use stock-photo-finder instead; do NOT use to fetch images via an API with attribution rules - use stock-photo-api instead.
What it loads into context
Frontmatter
203 tokensAlways available to the agent
SKILL.md
1,965 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: license record
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore9,340 bytesSKILL.mdcore9,340 bytesSKILL.mdcore9,340 bytesSKILL.mdcore9,340 bytesSKILL.mdcore9,340 bytesSKILL.mdcore9,340 bytes
Read SKILL.md
# Image License & Usage Rights
The license travels with the image, not with the website you found it on - the same photo can be free on one site and cost $500 on another, and the most common violation is running an editorial-use-only image in an ad. This skill decides whether a specific use is permitted, what the license obligates, and when a release is needed. It is the gate before anything commercial ships.
First principle: verify on the source page, never from memory. Licenses change and sites revise terms; open the image's actual license page on the provider before relying on any summary, including this one.
## Operating procedure
Order matters: the use classification (Step 2) determines which license terms even apply, so classify before reading.
### Step 1: Gather inputs
- The image: source URL, provider, and any license/asset ID.
- The intended use: where it will appear, for how long, and at what scale.impact-report-builder
When your agent uses it
Builds a nonprofit's annual or program impact report that funders and donors actually trust - leading with outcomes over activities, pairing every number with one human story, admitting shortfalls honestly, including a financial transparency block, and computing program cost per outcome. Use when a nonprofit leader says "write our annual report", "build our impact report", "a funder wants to see our results", "how do we report on the grant", or "what goes in a program report". Do NOT use for board meeting materials - use nonprofit-board-pack instead - and do NOT use for individual donor appeals or thank-yous - use donor-communications instead.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
2,096 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: report skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore9,495 bytesSKILL.mdcore9,495 bytes
Read SKILL.md
# Impact Report Builder
Most nonprofit impact reports are activity logs with photos: sessions delivered, events held, meals served - everything the organization did, nothing about what changed. Funders read hundreds of these and trust none, because a report that claims only wins and counts only activities is indistinguishable from marketing. This skill builds the report that earns renewals: outcomes first, numbers paired with stories, misses admitted, and the money accounted for.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students. The report's numbers are the pack's canonical outcomes from the logic model in grant-proposal-writer: 72 percent of 250 students gained at least one grade level in reading - the same figures donor-communications uses in stories and appeals. One organization, one set of numbers, everywhere.
## Operating procedure
### Step 1: Fix the audience and the number set
One report can serve foundation funders, individual donors, and the board with light re-framing, but the numbers must be one set. Pull the canonical outcomes (from the logic model), the year's financials (revenue by source, expenses by function), and last year's report if one exists. Any figure that cannot be sourced is labeled an estimate in the report itself - silent precision is how trust dies on page three.
### Step 2: Apply the outcomes-over-activities rule
inbox-zero
When your agent uses it
Processes any email backlog to zero using the 4Ds - Delete, Delegate, Defer, Do - with a mass-archive strategy for the obvious, a touch-each-email-once discipline, and a keep-it-clear system of batched processing windows, ruthless unsubscribing, filters, and a minimal folder setup. Use when someone says "I have 5,000 unread emails", "help me get to inbox zero", "email is eating my whole day", or treats their inbox as a to-do list. Do NOT use for drafting the reply emails themselves or prioritization rules for an ongoing support queue - use email-triage instead - or for protecting focus time around the email windows - use deep-work-planner instead.
What it loads into context
Frontmatter
164 tokensAlways available to the agent
SKILL.md
844 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The core idea
- The 4Ds
- Processing the backlog
- Keep it at zero
- Folder/label setup (minimal)
- A simple daily routine
- Anti-patterns
Files
SKILL.mdcore3,577 bytes
Read SKILL.md
# Inbox Zero
Inbox Zero isn't an empty inbox for its own sake - it's a clear mind. The inbox is a
processing queue, not a storage system or a to-do list. This skill clears it and keeps
it clear.
## The core idea
Your inbox should hold only **unprocessed** messages. Once you've looked at an email,
it should not stay in the inbox - it gets a decision. Touch each email once.
## The 4Ds
For every email, make one of four decisions immediately:index-advisor
When your agent uses it
Recommends, orders, and prunes indexes for a specific query or table - composite column order, selectivity rules, partial and covering indexes, duplicate/unused cleanup, and write-amplification tradeoffs. Use when a query is slow and EXPLAIN shows a Seq Scan or a sort, before adding a CREATE INDEX, or when auditing a table's index set for bloat or duplicates. Do NOT use to diagnose an unknown slow query from scratch - start with sql-query-optimizer; do NOT use when the query shape itself is the problem (function-wrapped predicates, leading wildcards, correlated subqueries) - use query-rewriter instead; do NOT use when the table is too large and needs partitioning or a time-series strategy - use partition-planner instead.
What it loads into context
Frontmatter
174 tokensAlways available to the agent
SKILL.md
1,689 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Workflow
- Worked example
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore7,231 bytes
Read SKILL.md
# Index Advisor
An index trades write speed and disk for read speed. Add the fewest indexes that serve real query shapes, order their columns deliberately, and prove every change with a plan. The costly mistake is the speculative index: it fails to help the read (wrong column order, predicate too unselective) while permanently taxing every write. This skill owns index design only - plan diagnosis belongs to sql-query-optimizer, and query restructuring belongs to query-rewriter.
## Inputs to collect
1. The slow query text and its `EXPLAIN (ANALYZE, BUFFERS)` output (Postgres) or `EXPLAIN ANALYZE` / `EXPLAIN FORMAT=JSON` (MySQL). No plan, no recommendation.
2. The table's row count and the existing index list (`\d table` / `SHOW INDEX`).
3. Predicate selectivity: roughly what fraction of rows each WHERE predicate matches. Estimate from `n_distinct` in `pg_stats` or a quick `COUNT(*)` ratio; label estimates as guesses.
4. The write profile: is this table write-hot (high INSERT/UPDATE rate), and which columns get updated?
5. For an audit: `pg_stat_user_indexes` (or the MySQL equivalent) over a representative window.
## Workflow
injury-prehab
When your agent uses it
Designs targeted prehab routines - sets, reps, and weekly placement - that build resilience at the four most common failure zones: shoulders, knees, lower back, and hips, matched to the user's training program. Use when someone says "my shoulders feel beat up from bench", "how do I bulletproof my knees for running", "prevent lower back tweaks from deadlifts", or is starting a new training block. General movement guidance for injury PREVENTION, not medical advice or physical therapy - existing injuries and pain go to a licensed physio. Do NOT use for a general desk-worker flexibility routine - use mobility-routine instead; for the main resistance program itself, use strength-training-plan; for conditioning, use fitness-program.
What it loads into context
Frontmatter
168 tokensAlways available to the agent
SKILL.md
1,789 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,671 bytes
Read SKILL.md
# Injury Prehab
Most training injuries announce themselves weeks in advance - a cranky shoulder, a knee that aches on stairs - and the cheap fix is targeted resilience work done before failure, not rehab after it. This skill programs that work for the four zones where lifters and runners actually break: shoulders, knees, lower back, and hips. It prevents; it does not treat. Anything that already hurts belongs with a physiotherapist.
## Operating procedure
### Step 1: Gather inputs
Collect: the current training program (lifting split, running volume, sport) so prehab can attach to it; training days per week; the zones that feel tight, weak, or historically troublesome; any current pain or past injury (past injury means bias volume to that zone; current pain means stop - see Escalation); available equipment (bands cover 90% of prehab); and minutes willing to spend (default 10-15 before sessions). If the user cannot name a weak zone, treat all four as guesses and start with the balanced circuit.
### Step 2: Screen before programming
Sharp pain, radiating pain, numbness or tingling, joint swelling or locking, or night pain disqualify a zone from prehab programming - those are physio referrals, not exercise-selection problems. Prehab loads are light, but loading an undiagnosed injury still makes it worse.
instagram-post-builder
When your agent uses it
Interactively builds a complete, copy-paste-ready Instagram post package - three hook variants using distinct patterns, a caption body, one clear CTA, a 15-20 hashtag set across reach tiers, alt text, and a visual direction with shot list or slide breakdown - through a one-question-at-a-time interview, optionally grounded in a trend-educator briefing. Use when someone says "write an Instagram post for my launch", "build the full post package for this reel", "put together a carousel with hooks and hashtags", or reaches the build stage after trend-educator. Do NOT use for finding what is trending - use instagram-trend-scout instead; for a single caption without the full package - use social-caption-writer; for planning a month of content - use social-content-calendar.
What it loads into context
Frontmatter
222 tokensAlways available to the agent
SKILL.md
2,525 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Input Schema
- Workflow
- Output Schema
- Output Format
- Error Handling
- Do NOT
- Quality bar
- Flywheel Connections
Files
SKILL.mdcore10,461 bytes
Read SKILL.md
# Instagram Post Builder
**Stage: S3 - Build** (final skill in the Instagram Trend Post Builder flywheel)
This is where trend awareness becomes an actual post. It takes the trend card from `trend-educator` as *fuel*, but the post is built around the user's own message - the trend informs the packaging, it does not dictate the content. The result is a copy-paste-ready package.
The defining behavior of this skill is **conversational, one question at a time**. Do not interrogate the user with a form. Ask, listen, then ask the next thing - adapting to what they said.
## Input Schema
| Field | Type | Required | Default | Description |
|-------|------|----------|---------|-------------|
| `user_message` | string | yes* | - | What the user wants to say / the topic of the post. *If absent, the skill elicits it as the first question rather than failing. |
| `educator_brief` | object | no | - | The output from `trend-educator` (`trend_summary`, `why_it_works`, `opportunity_angles`, `confidence_note`). When present, the build is grounded in these trends. When absent, the skill still works in standalone mode. |instagram-trend-scout
When your agent uses it
Researches what is currently working on Instagram for a niche using web search only - no API or login - and returns a structured scouting report of top content, winning Reel and carousel formats, reusable hook patterns, content gaps, and an engagement benchmark with honest confidence labels. Use when someone asks "what's trending on Instagram for my niche", "scout Reels formats for home baristas", "what hooks are working right now", or is starting a weekly content cycle and wants signals before creating. Do NOT use for explaining why trends work in plain language - use trend-educator instead; for drafting the actual post, use instagram-post-builder.
What it loads into context
Frontmatter
196 tokensAlways available to the agent
SKILL.md
2,097 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Input Schema
- Workflow
- Output Schema
- Output Format
- Error Handling
- Flywheel Connections
Files
SKILL.mdcore8,738 bytesSKILL.mdcore8,738 bytesSKILL.mdcore8,738 bytesSKILL.mdcore8,738 bytes
Read SKILL.md
# Instagram Trend Scout
**Stage: S1 - Research** (first skill in the Instagram Trend Post Builder flywheel)
This skill answers one question: *what is working on Instagram for this niche, right now?* It gathers public signals through web search - it does not log into Instagram or call any private API - and returns a structured, machine-readable scouting report that the next skill (`trend-educator`) turns into a plain-language briefing.
Be honest about confidence. Web search surfaces secondhand signals (roundup articles, creator breakdowns, public engagement screenshots), not the Instagram Graph API. Label every number as an estimate and never invent precise metrics.
## Input Schema
| Field | Type | Required | Default | Description |
|-------|------|----------|---------|-------------|
| `keyword` | string | yes | - | The niche, topic, or audience to scout (e.g. "home barista", "B2B SaaS founders", "postpartum fitness"). |
| `time_range` | string | no | `"7d"` | How fresh the trends must be. One of `"7d"`, `"14d"`, `"30d"`, `"90d"`. Controls how aggressively to bias search queries toward recency. |internal-linking-mapper
When your agent uses it
Maps internal links and anchor text between a new or updated page and an existing URL inventory to strengthen a topic cluster, producing a source/target/anchor link table plus a hub-and-spoke check. Use when someone asks "which pages should this article link to", "how do I internally link my new post", "is this page orphaned", or has a draft plus a sitemap, URL export, or pillar/cluster map and needs to decide the inbound links, outbound links, and anchors. Do NOT use to design the cluster or pillar architecture from a keyword export - use keyword-cluster-builder instead; do NOT use for on-page rewrites of titles, headings, or body copy - use seo-optimizer instead; do NOT use when a page is slipping in rankings and needs SERP-intent update actions - use content-refresh-auditor instead.
What it loads into context
Frontmatter
192 tokensAlways available to the agent
SKILL.md
1,505 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: hub-and-spoke link map
- Deliverable
- Do NOT
- Quality bar
- Neighbors
Files
SKILL.mdcore6,831 bytesSKILL.mdcore6,831 bytes
Read SKILL.md
# Internal Linking Mapper
Wire one page into an existing site so the whole topic cluster gains authority: pick the outbound links from the new page, the inbound links into it, and descriptive anchor text for each. The costly mistake this prevents is the orphan - a page with zero inbound internal links that search engines rarely crawl and never rank - and its opposite, exact-match anchor spam that reads as manipulation. This skill assumes the cluster already exists and the page content is set; it only decides the links.
## Operating procedure
Order matters: you cannot plan links until you know the inventory, and you cannot write anchors until you know the links.
### Step 1: Gather inputs
- The target page: topic, primary keyword, and draft or URL.
- The existing URL inventory: sitemap, URL export, or pillar/cluster map. If none is supplied, ask for one - never invent URLs.
- Optional but valuable: which page is the top money/conversion page, and any pages known to be thin, noindex, or non-canonical (these are link dead-ends).
interview-debrief-synthesizer
When your agent uses it
Consolidates multiple interviewer scorecards into a calibrated hire/no-hire summary that foregrounds evidence, surfaces disagreement, and flags bias risks for a human panel. Use when you have collected scorecards or written feedback from a completed interview loop and need to run the debrief, calibrate divergent ratings, or produce a defensible decision write-up. Do NOT use when you are preparing questions or rubrics before interviews - use interview-question-kit instead.
What it loads into context
Frontmatter
114 tokensAlways available to the agent
SKILL.md
930 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,641 bytes
Read SKILL.md
# Interview Debrief Synthesizer
Turn a set of independent interviewer scorecards into one calibrated, evidence-led decision summary that a human panel can defend.
## Workflow
1. **Gather inputs.** Collect every interviewer's scorecard or written feedback and the role's competency rubric with its weights. If the rubric is missing, ask for it; do not infer a bar from the comments. Push to run the synthesis within 48 hours of the final interview - recall of specific candidate answers decays fast, and evidence gathered late drifts toward impression.
2. **Aggregate against the shared rubric.** Map each interviewer's ratings to the defined competencies. Present per-competency scores side by side with each interviewer's cited evidence. Compute the weighted picture, but label it explicitly as input to discussion, never a verdict. Flag any competency no interviewer actually assessed, and treat any competency assessed by only one interviewer as thin coverage worth naming.
3. **Surface disagreement instead of averaging it.** Where interviewers diverge by 2 or more points on the same competency, show both ratings with both sides' evidence. Treat divergence as a signal of missing information or different bars, not a number to split; recommend the panel discuss or gather more signal rather than collapsing a 2 and a 4 into a 3. Divergence of a single point is normal calibration noise - do not manufacture a controversy from it.
4. **Separate evidence from impression.** For each claim, distinguish observed behavior ("could not explain their own design tradeoffs") from impression ("seemed junior"). Down-weight or set aside comments with no behavior behind them, and note which interviewer should supply the missing evidence.
5. **Flag bias and process risks.** Mark untethered "fit" or "polish" comments, similarity bias ("reminds me of us"), remarks about communication style that may track accent or culture, harsher scoring on identical evidence, and any reference to a protected characteristic. Frame these as risks for the panel to examine, not as accusations.
6. **Frame the decision and its rationale.** Summarize strengths, concerns, and the open questions the decision hinges on, mapped to the leveling bar. Present the options - hire / no-hire / more signal needed - each with its supporting evidence. Capture the final rationale in writing for consistency and auditability.
## Deliverable
interview-guide-builder
When your agent uses it
Writes a non-leading user-research discussion guide with warm-up flow, open core questions, an attached probe bank, time budgets, and a pre-field bias audit. Use when someone asks "write my interview guide", "what should I ask users in discovery interviews", "review my questions for leading language", or is preparing generative or evaluative user interviews. Do NOT use for hiring-loop questions with competencies and scoring anchors - use interview-question-kit instead. Do NOT use for sourcing deep domain knowledge from subject-matter experts with laddering - use expert-interview instead. To analyze the notes afterward, use interview-synthesis.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,360 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,354 bytes
Read SKILL.md
# Interview Guide Builder
A discussion guide is scaffolding, not a script: it keeps the researcher oriented while leaving room for the participant to lead. The costly failure it prevents is manufactured confirmation - a guide full of leading questions returns exactly the answers the team already believed, and the product ships into a market that was never actually heard.
## Operating procedure
### Step 1: Gather inputs
Collect before drafting:
- Research goal in one sentence, and whether the round is **generative** (understand behavior and needs) or **evaluative** (react to a concept or prototype). Default: generative.
- The 3-5 things the team must learn - these become topic blocks; anything beyond five gets cut or deferred to another round.
- Session length (default 60 minutes), participant profile, and whether sessions are recorded.
- The team's current hypotheses - written down explicitly so they can be kept OUT of the question stems.interview-question-kit
When your agent uses it
Generates a structured candidate-interview question set where every behavioral (STAR) and role-specific item maps to a defined competency and ships with strong/borderline/red-flag answer anchors plus delivery notes. Use when someone asks "write interview questions for this role", "build an interview loop", "turn this rubric into questions per interviewer", or is planning a hiring interview, building a question bank, or standardizing how a panel assesses candidates. Do NOT use to consolidate completed scorecards into a hire decision - use interview-debrief-synthesizer instead; do NOT use for user-research or customer discovery interviews - use interview-guide-builder instead; do NOT use to build the competency rubric or scorecard itself - use hiring-scorecard or screening-rubric-builder instead.
What it loads into context
Frontmatter
171 tokensAlways available to the agent
SKILL.md
1,589 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: one kit entry (copy this structure per item)
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore7,598 bytes
Read SKILL.md
# Interview Question Kit
Turn a role's competencies into a fair, comparable question set with per-question scoring anchors, so interviewers assess evidence instead of rapport. The costly mistake this prevents is the unstructured loop: five interviewers improvising different questions, scoring on likability, producing scorecards that cannot be compared - which is how teams hire the best storyteller instead of the best candidate, and how legally risky questions slip in.
## Operating procedure
Order matters: questions written before the competency map exist to sound clever, not to gather evidence.
### Step 1: Gather the rubric
Collect the role's defined, job-relevant competencies and their weights, plus the job description, loop size (how many interviewers, how long each session), and seniority level. If no rubric exists, derive 4-6 competencies from the job description, label them as derived, and confirm them before writing a single question. Never write a question that does not map to a competency on this list.
### Step 2: Allocate competencies across the loop
interview-synthesis
When your agent uses it
Turns raw interview notes into themed, evidence-tagged insights via a two-pass affinity-mapping procedure, with frequency and confidence ratings and stakeholder-ready quotes. Use when someone asks "synthesize these interviews", "what themes came out of our research round", "turn these transcripts into findings", or has finished a qualitative round and needs actionable output. Do NOT use for synthesizing published papers or mixed documents - use research-synthesis instead. Do NOT use for building the personas themselves - use user-persona instead. Do NOT use for consolidating hiring-interview scorecards - use interview-debrief-synthesizer instead.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,408 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,964 bytes
Read SKILL.md
# Interview Synthesis
Synthesis is where data becomes insight. The job is not to summarize what people said - it is to identify the underlying pattern that explains why they said it and what it means for design or strategy. The costly failure it prevents is the highlight reel: a deck of cherry-picked quotes that confirms what the team already believed, laundered through the word "research."
## Operating procedure
### Step 1: Gather inputs and check the dataset
Collect: the transcripts or structured notes, the original research questions, and who the findings are for (design team vs. executives changes the framing, not the rigor). Enforce the data rules:
- Every interview needs a cleaned transcript or structured notes with participant ID, date, and verbatim quotes marked. Do not synthesize from memory or from the researcher's paraphrase alone.
- **Minimum viable dataset: 5 interviews** for foundational themes. Fewer is workable only with an explicit low-confidence label on every finding.
- **Check for saturation**: if the last 3 interviews produced no new themes - only more evidence for existing ones - the round has reached saturation and the theme set can be treated as stable. If new themes were still appearing in the final interviews, say so and flag that more interviews would likely surface more; the synthesis is a snapshot, not the territory.
investment-basics
When your agent uses it
Teaches personal investing fundamentals - readiness checks, asset allocation, diversification with low-cost index funds, fee awareness, tax-advantaged account ordering, and rebalancing - and produces a written starter plan. Use when someone asks "how do I start investing", "what should my stock/bond split be", "are index funds enough", or "should I invest or pay off debt first". General financial education, not personalized investment advice. Do NOT use for a full multi-goal financial plan with budgeting and insurance - use financial-planner instead; do NOT use for modeling when someone can retire - use retirement-projection instead; for building the cash cushion first, see emergency-fund-planner.
What it loads into context
Frontmatter
157 tokensAlways available to the agent
SKILL.md
1,730 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: starter plan
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,952 bytes
Read SKILL.md
# Investment Basics
The costliest beginner mistakes are not bad fund picks - they are investing before the foundation exists, paying 1%+ in fees for decades, concentrating in single stocks or employer stock, and panic-selling in the first real downturn. This skill produces a boring, durable starter plan and a written allocation the user can hold through a crash.
This is general financial education, not personalized investment, tax, or legal advice. Markets carry risk and losses are possible. Account types, contribution limits, and tax rules vary by country - always verify local rules.
## Operating procedure
Order matters: readiness gates come before allocation, and allocation comes before any fund discussion, because allocation drives the large majority of a portfolio's risk and return - fund selection is a rounding error next to it.
### Step 1: Gather inputs
Collect, with defaults where the user is unsure (label guesses as guesses):
investor-pipeline-crm
When your agent uses it
Use when a founder is managing an active raise across many investors. Triggers on "track my fundraise", "investor pipeline", "fundraising CRM", "how to create FOMO", "run my raise as a process", "investor follow-up cadence", "how long should fundraising take". Turns a scattered raise into a managed, time-boxed process.
What it loads into context
Frontmatter
84 tokensAlways available to the agent
SKILL.md
728 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- Pipeline stages
- Time-box the raise
- Manufacture momentum honestly
- Cadence and follow-up
- Anti-patterns
- Deliverable
Files
SKILL.mdcore3,195 bytes
Read SKILL.md
# Investor Pipeline CRM
A raise that drifts dies. Investors are pattern-matchers; a process that wanders
for months signals a company nobody else wants. The antidote is to run the raise
like a disciplined sales pipeline, time-boxed to manufacture a decision window.
## When to use this skill
Use it once [[investor-targeting]] produces the list and outreach begins. It is
the operating system for the live raise, from first email to signed term sheet
([[term-sheet-negotiation]]).
## Pipeline stages
investor-targeting
When your agent uses it
Use when a founder needs to decide which investors to approach and how to reach them. Triggers on "which VCs should I target", "build my investor list", "find the right investors", "warm intro path", "angels vs VCs", "who invests in my space", "investor research". Produces a ranked, researched target list with warm paths.
What it loads into context
Frontmatter
84 tokensAlways available to the agent
SKILL.md
990 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- Qualify on fit, not fame
- Angels vs VCs vs other capital
- Map the warm path
- Sequence the list
- Anti-patterns
- Quality bar
- Deliverable
Files
SKILL.mdcore4,144 bytes
Read SKILL.md
# Investor Targeting
A raise is a sales process, and a sales process with the wrong leads fails no
matter how good the pitch. Most founders spray a generic list and wonder why
conversion is low. This skill builds a tight, researched, warm-path-first list.
## When to use this skill
Use it once [[fundraising-stage-selector]] fixes the stage and check size, and
before [[vc-pitch-email]] writes outreach. Feed the output into
[[investor-pipeline-crm]] to run the process.
## Qualify on fit, not fame
investor-update-writer
When your agent uses it
Drafts a monthly or quarterly investor update - TL;DR, consistent metric block (revenue, growth rate, burn, runway), highlights, honest lowlights, a specific ask, and next-period priorities - in a repeatable one-screen format. Use when someone asks "write my investor update", "what should go in my monthly update", "how do I tell investors about a bad month", or when a founder is due to report to their cap table. Do NOT use for updates to internal teams, executives, or non-investor stakeholders - use stakeholder-update instead. Do NOT use for crafting the fundraising story for new investors - use fundraising-narrative instead.
What it loads into context
Frontmatter
153 tokensAlways available to the agent
SKILL.md
1,604 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template: monthly update
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,858 bytes
Read SKILL.md
# Investor Update Writer
Investor updates compound: a founder who sends an honest, consistent update every month builds the trust that makes the next round, the hard intro, and the bad-news conversation easier. The costly failure this skill prevents is the silent founder - updates that stop when metrics dip, which investors read (correctly) as a distress signal, and updates so padded with hype that the one real ask gets skimmed past.
## Operating procedure
Write in this order because the TL;DR is a compression of everything else - draft it last even though it appears first.
### Step 1: Gather inputs
Collect before drafting. Label estimates as estimates in the update itself.
1. Cadence and audience: monthly is the default for pre-Series B; quarterly acceptable after. Same recipients every time.
2. The metric set (see Step 2) with current and prior-period values.jd-bias-scrubber
When your agent uses it
Audits a job description for exclusionary, gendered, age-coded, ableist, and pedigree-gatekeeping language and returns a findings table with neutral rewrites and severity. Use when drafting, reviewing, or auditing a job posting or JD before it is published, when asked to check a role description for biased or non-inclusive wording, or before a requisition goes to a job board.
What it loads into context
Frontmatter
101 tokensAlways available to the agent
SKILL.md
895 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,213 bytes
Read SKILL.md
# JD Bias Scrubber
Surface words and requirements in a job description that deter qualified candidates without adding predictive value, and propose neutral rewrites the hiring manager can accept or reject. You flag; you never silently rewrite intent. Remember the asymmetry the audit exists to correct: a widely cited internal HP study found women tend to apply only when they meet ~100% of listed requirements while men apply at ~60% - every unnecessary requirement shrinks the pool unevenly.
## Workflow
1. **Scan for gender-coded language.** Flag masculine-coded terms ("dominant," "aggressive," "rockstar," "crush it," "competitive," "ninja") and feminine-coded terms that skew the pool; propose role-relevant neutral phrasing ("sets a high bar," "delivers results"). Flag pronoun assumptions and replace with "you" or the role title.
2. **Catch age and experience coding.** Flag "digital native," "recent grad," "high energy," "young team," and maximums like "no more than 5 years." Question any years-of-experience minimum above what the work requires - minimums above 5 years rarely map to competency for non-executive roles; convert them to the actual competency or a soft range, since high minimums screen out career changers and returners.
3. **Surface ableist and physical-demand language.** Flag "able-bodied," blanket "must stand for hours," and "fast-paced high-pressure" used as a personality filter. Flag metaphors ("see," "walk," "strong") when not literal job functions. Where a physical demand is genuine, state it precisely and pair it with an accommodation note instead of a blanket exclusion.
4. **Question credential and pedigree gatekeeping.** Flag degree requirements, "top university," "Big Tech experience," and citizenship/native-speaker phrasing that exceed legal or job need. Suggest the demonstrable skill instead. Flag "native English speaker" as possible national-origin discrimination; prefer "fluent professional English." Flag a required-qualifications list longer than 5-7 items: recommend demoting the rest to "nice to have."
5. **Output the findings table.** Return one row per flagged phrase with columns: flagged phrase, why it may exclude, suggested neutral alternative, severity (legal-risk / pool-shrinking / tone). Preserve substantive requirements; challenge only those with no job-relevant basis.
## Deliverable
Produce the findings table - one row per flagged phrase with the phrase, why it may exclude, a usable neutral rewrite, and severity - topped by a two-line summary: total flags by severity, and whether the JD is publishable once legal-risk rewrites are accepted or explicitly waived by counsel.jetpack-compose-builder
When your agent uses it
Builds Android Jetpack Compose UI with hoisted state, unidirectional data flow, and recomposition-safe patterns backed by measured stability rules. Use when someone asks "build this screen in Compose", "where should this state live", "why does my LazyColumn stutter", or is writing Composable functions, ViewModel and StateFlow wiring, or fixing recomposition jank in Kotlin. Do NOT use when the screen targets Flutter - use flutter-widget-architect instead; for iOS declarative UI use swift-ui; for React Native screens use react-native-pro; and if jank persists after state fixes, profile with mobile-perf-profiler.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,944 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Recomposition cost rules
- Worked artifact: bad vs good list row
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore9,032 bytes
Read SKILL.md
# Jetpack Compose Builder
Build the state and data flow first, the pixels second - most Compose bugs are state-ownership and recomposition bugs, not layout bugs. The costly failure is a screen that works in a demo, then recomposes entire subtrees on every keystroke or scroll frame because state lives in the wrong place and parameters are unstable.
## Operating procedure
### Step 1: Gather inputs
Before writing a composable, establish:
1. The screen's state shape: what the user sees, what can change, and what triggers each change.
2. Which state is UI-local (text field focus, expanded flags) versus screen-level (loaded data, selection) versus app-level (auth, theme).
3. The data source: ViewModel exposing StateFlow, or a snapshot-state holder for pure-UI screens.
4. List characteristics if any: item count, item identity (stable server ids?), reorder/insert behavior.
5. Refresh-rate target of the flagship devices in the install base (60Hz gives a 16.6ms frame budget, 120Hz gives 8.3ms).jira-ticket-writer
When your agent uses it
Writes ready-to-build Jira stories with user-story summaries, Given/When/Then acceptance criteria, subtasks, and story-point guidance, checked against INVEST and a Definition of Ready. Use when someone asks "write a Jira ticket for this", "turn these meeting notes into a story", "are these acceptance criteria testable", "should I split this story", or "how many points is this". Do NOT use for configuring Linear teams, cycles, labels, or triage workflows - use linear-workflow instead. For deciding what fits in the sprint, use sprint-planning.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
1,928 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: bad vs good
- Deliverable
- Do NOT
- Quality bar
- Related skills
Files
SKILL.mdcore8,440 bytesSKILL.mdcore8,440 bytes
Read SKILL.md
# Jira Ticket Writer
A good ticket is understandable by someone who was not in the conversation. The costly failure mode is the "fix the thing" ticket: it looks fine the day it is written, then a week later nobody can build it, estimate it, or verify it, and the sprint absorbs a re-discovery meeting that the ticket was supposed to replace. This skill produces stories that are ready to estimate and build on first read.
## Operating procedure
Follow the steps in order: the summary shapes the acceptance criteria, the criteria expose the edge cases, and the edge cases decide whether the story needs splitting. Estimating before the criteria exist produces numbers nobody can defend.
### Step 1: Gather inputs
Collect these before drafting. Where the requester does not know, record the best guess and label it a guess.
1. Who the user is (a real role - "billing admin", not "user").
2. The problem and why now - the source conversation, incident, or request to link.job-application
When your agent uses it
Tailors a resume and writes a cover letter for one specific job description - mirroring JD keywords for ATS parsing, reordering bullets by relevance, quantifying impact, and reporting which JD keywords matched and which are missing. Use when someone asks "tailor my resume to this job", "write a cover letter for this posting", "will my resume pass the ATS", or shares a job description alongside their resume. Do NOT use for writing or restructuring a resume without a target job - use resume-writer instead; for critique of an existing resume, use resume-reviewer.
What it loads into context
Frontmatter
133 tokensAlways available to the agent
SKILL.md
717 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- Resume tailoring
- ATS rules
- Cover letter (3 short paragraphs)
- Honesty guardrails
- Anti-patterns
- Output
Files
SKILL.mdcore3,299 bytes
Read SKILL.md
# Job Application Writer
You adapt a candidate's resume and write a cover letter targeted to one job description, balancing ATS keyword matching with genuine, human-readable impact.
## Process
1. Collect: the full job description, the candidate's current resume, and any extra context (target salary, why they want it).
2. Extract from the JD: required skills, preferred skills, the exact keywords/titles used, and the top 3 responsibilities.
3. Map the candidate's experience to those, prioritizing overlaps.
4. Rewrite bullets and draft a cover letter.
## Resume tailoring
- **Mirror the JD's language.** If they say "stakeholder management," use that exact phrase where true - ATS matches strings, not synonyms.job-description-writer
When your agent uses it
Draft an inclusive, role-relevant, level-calibrated job posting from a hiring manager's role brief. Use when an employer is writing or revising a job description, JD, or job posting - you have a role brief, leveling rubric, or comp band and need outcome-based requirements, plain compensation/logistics, and inclusive tone. Do NOT use when tailoring a candidate's resume or cover letter to a posting - use job-application instead.
What it loads into context
Frontmatter
109 tokensAlways available to the agent
SKILL.md
957 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,579 bytes
Read SKILL.md
# Job Description Writer
Turn an employer's role brief into a posting a strong candidate reads to the end and a fair process can be built on. A JD is a hiring instrument, not a wish list: every line either attracts the right candidate and predicts on-the-job success, or it narrows the pool for no reason.
## Workflow
1. Confirm the role brief and level. Pull the role's leveling rubric (scope, autonomy, ambiguity, influence) and map the title to one level. If the level is missing or the brief is vague, ask before drafting - a title without a level produces mismatched applicants and inconsistent interviews.
2. Write 4-6 outcome statements, not task lists. State what this person must accomplish in 12 months ("Ship the billing migration with zero revenue-impacting incidents"), not activities ("Write code"). Outcomes make the bar legible and force the hiring manager to clarify what they are buying.
3. Separate must-haves from bonus. Cap must-haves at 5-7 and include only what is genuinely needed on day one. Move everything else to "bonus." Replace proxy credentials ("top-tier CS degree") with the underlying competency ("can reason about distributed-systems tradeoffs").
4. State experience as a signal, not a gate. Frame years of experience as a soft signal, never a hard requirement. Never set a maximum years-of-experience or any ceiling that screens for age.
5. State compensation and logistics plainly. Include a real pay range pulled from the employer's comp bands or market data, plus location/remote policy and travel. A posted range whose top is more than roughly 50% above its bottom reads as evasive and depresses applications; if the honest band is that wide, the role is probably spanning two levels - split it or narrow it. Never invent a range from intuition; if no band is provided, flag it and route to whoever owns comp.
6. Set human, inclusive tone. Use plain second-person language. Cut jargon, superlatives, and culture-fit code ("work hard play hard," "ninja," "young and energetic"). State concrete inclusion and accommodation commitments. Describe the team and the work, not a personality the candidate must perform. Keep the whole posting in the 300-700 word range - long enough to be concrete, short enough that candidates read to the compensation line.
7. Hand off for ownership. You draft; a human hiring manager and recruiter own the final posting and any legal review. Flag any requirement you cannot tie to job performance, and route comp ranges and EEO language to the people who own them.
## Deliverablejob-interview-prep
When your agent uses it
Prepares a candidate for a specific job interview - company research checklist, a STAR story bank mined from their resume and mapped to common competencies, a questions-to-ask bank, salary-question deflection lines, and a mock-interview loop. Use when a user says "I have an interview next week, help me prepare", "help me practice answers for this interview", "what should I ask the interviewer", "build me STAR stories from my resume", or "they're going to ask about salary, what do I say". Do NOT use for negotiating after an offer arrives - use salary-negotiation instead - and do NOT use for fixing the resume itself - use resume-reviewer instead.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
2,124 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Concrete thresholds
- Deliverable
- Do NOT
- Quality bar
- Escalation and neighbors
Files
SKILL.mdcore9,526 bytes
Read SKILL.md
# Job Interview Prep
Interviews are won by candidates who show up with specific, rehearsed evidence and lost by candidates who improvise answers to questions they could have predicted. The costly mistake this skill prevents is walking in with a strong resume and no story bank - freezing on "tell me about a time you handled conflict," rambling for four minutes, and never landing the result. Preparation converts what the candidate already did into answers that land in ninety seconds.
## Operating procedure
Run the steps in order. Story mining (Step 3) comes before question practice, because every behavioral answer draws from the same bank - building the bank once beats improvising per question. The mock loop (Step 6) comes last because rehearsal without material rehearses rambling.
### Step 1: Collect the inputs
- **Resume** (required) - the raw material for the story bank.
- **Job posting** (required) - determines which competencies to weight.
- **Interview format** (required) - recruiter screen, hiring-manager, panel, technical, final round. Default if unknown: assume a hiring-manager behavioral interview and label that a guess.
- **Company name and interviewer names** (as available) - feeds research and questions-to-ask.journal-framework
When your agent uses it
Designs a sustainable daily journaling practice - morning pages, gratitude, intention setting, evening reflection, and weekly review - matched to the user's goal, with copy-paste templates and a rotating prompt library. Use when someone asks "help me start journaling", "give me journal prompts", "I keep abandoning my journal", or "build me a morning pages routine". Do NOT use for building a general habit or streak system - use habit-builder instead; do NOT use for organizing notes and knowledge into a retrieval system - use second-brain instead; for structured end-of-week planning reviews, see weekly-review.
What it loads into context
Frontmatter
137 tokensAlways available to the agent
SKILL.md
1,616 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: weekly template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,173 bytes
Read SKILL.md
# Journal Framework
Most journaling practices die within two weeks, and they die for predictable reasons: the ritual is too long, it has no anchor to an existing habit, and the prompts go stale. The fix is not more willpower - it is a practice sized to the user's actual life, with a tiny minimum, a fixed anchor, and prompts matched to what they want out of it.
Journaling supports general wellbeing but is not a substitute for therapy or mental-health care. If the user describes persistent distress, self-harm thoughts, or symptoms interfering with daily life, recommend a licensed professional before continuing.
## Operating procedure
Order matters: the goal determines the prompts, and the time budget determines how many components survive - designing components before knowing the budget produces a routine that gets abandoned.
### Step 1: Gather inputs
Collect, with defaults:
jtbd-extractor
When your agent uses it
Extracts Jobs To Be Done from qualitative research data - job stories in the "When I…, I want to…, so I can…" form, a forces-of-progress map (push, pull, anxiety, habit), and functional, emotional, and social dimensions, each backed by verbatim quotes. Use when someone says "extract the jobs from these interviews", "run a JTBD analysis on this diary study", "why do users switch to or from us", or has transcripts and wants solution-agnostic goals rather than feature requests. Do NOT use for clustering findings into themes across a study - use interview-synthesis instead; for building persona profiles from research, use user-persona; for planning the interviews themselves, use interview-guide-builder.
What it loads into context
Frontmatter
168 tokensAlways available to the agent
SKILL.md
1,424 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked example
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore6,735 bytes
Read SKILL.md
# JTBD Extractor
Jobs To Be Done reframes research around the progress people are trying to make, not product features or user attributes - the Intercom/Ulwick lineage, not a cargo-culted template. The costly failure this skill prevents is shipping a "jobs inventory" that is really a feature wishlist with JTBD grammar pasted on: jobs contaminated with solution language are unstable, and a roadmap built on them chases tools instead of progress.
## Inputs to collect
1. The raw data: interview transcripts, diary-study entries, support tickets, or switch-moment notes. Verbatim material is required - summaries strip the quotes that serve as evidence. If only summaries exist, proceed but downgrade all evidence strength and say so.
2. Sample size and who was interviewed. Fewer than 5 participants means the inventory is directional; label it that way.
3. The decision the jobs will inform (roadmap, positioning, onboarding redesign). This sets granularity - positioning wants the main job; roadmap wants the sub-jobs.
4. Any existing jobs inventory to reconcile against, so new jobs merge rather than duplicate.
## Operating procedure
Order matters: forces are mapped per candidate job, so candidates must be isolated first; dimensions come last because they need the accumulated quote evidence.kafka-pipelines
When your agent uses it
Designs Kafka streaming pipelines end to end - partition-count sizing math, key choice, consumer-group sizing, offset strategy, delivery semantics, poll tuning, and dead-letter handling - with production configs. Use when someone asks "how many partitions should this topic have", "my consumer group keeps rebalancing", "consumer lag keeps growing", "how do I handle poison messages", or is designing a new topic or event-driven pipeline. Do NOT use for Spark batch or PySpark job tuning - use spark-jobs instead; do NOT use for HTTP webhook ingestion reliability - use webhook-receiver-hardener instead.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,844 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: partitioning decision
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,326 bytesSKILL.mdcore8,326 bytes
Read SKILL.md
# Kafka Pipelines
Partition count is the one Kafka decision that is nearly permanent: it caps consumer parallelism forever, and increasing it later rehashes keys and breaks per-key ordering for every keyed consumer. The two costly mistakes this skill prevents are under-partitioning a topic that later gets hot, and committing offsets before processing - which silently loses records on the first crash.
## Operating procedure
### Step 1: Gather inputs
Collect these before choosing any number. Label estimates as guesses and revisit after load testing.
1. Peak produce throughput (MB/s and messages/s) - peak, not average. Default guess: 3x the average.
2. Average message size.
3. Per-record consumer processing time, including downstream calls (DB writes, HTTP). This usually dominates.
4. Ordering requirement: which records must stay ordered relative to each other? That defines the key.kb-article-writer
When your agent uses it
Turns a resolved support ticket into a clear, searchable knowledge-base or help center article. Use when someone asks "write a KB article from this resolved ticket", "turn these resolved tickets into help center articles", or "write an internal knowledge base article for our support agents" - converting ticket resolutions into self-service documentation that deflects future tickets. Do NOT use for writing user-facing how-to guides from scratch - use help-documentation instead.
What it loads into context
Frontmatter
109 tokensAlways available to the agent
SKILL.md
963 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to create an article
- Article structure
- Worked example
- Writing rules
- SEO and findability
- Review and freshness
- Deliverable
- Quality bar
- What to avoid
Files
SKILL.mdcore4,769 bytes
Read SKILL.md
# Knowledge Base Article Writer
Every ticket that gets answered manually is a ticket that could have been self-served. A good knowledge-base article is written for the customer who is about to open a ticket, not for the agent who already knows the answer.
## When to create an article
Create an article when a ticket has been answered by two or more different customers in the past 60 days, or when a support lead flags a topic as recurring. Do not create articles for one-off edge cases or deprecated features.
## Article structure
Use this structure for every article:
1. Title: a plain-language question or task the customer is trying to accomplish ('How to reset your password', 'Why was I charged twice'). Use the exact words a customer would type into a search box.
2. Summary (2-3 sentences): what this article covers and who it is for.
3. Prerequisites (if applicable): what the customer needs before starting.keyword-cluster-builder
When your agent uses it
Collapses a raw keyword list into topic clusters and a pillar/cluster page architecture with one primary keyword and a prioritized build order per cluster. Use when you have a keyword export (CSV/GSC/keyword-tool dump) and need a content plan that says which pages to build, not a flat list. Do NOT use when labeling individual queries by intent or recommending page types - use search-intent-classifier instead; do NOT use when choosing internal links or anchor text between pages - use internal-linking-mapper instead.
What it loads into context
Frontmatter
123 tokensAlways available to the agent
SKILL.md
984 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Worked example
- Quality bar
- Do NOT
Files
SKILL.mdcore4,759 bytes
Read SKILL.md
# Keyword Cluster Builder
Turn hundreds of queries into a small set of clusters, each earning ONE page, arranged so topical authority compounds. A keyword list is not a content plan: one thin page per keyword cannibalizes your own rankings and wastes crawl budget.
## Workflow
1. **Dedupe and discard.** Drop near-zero-volume singletons (under ~10 searches/month with no transactional intent), branded/navigational queries you cannot rank for, and queries whose SERP is pure ads/maps/widgets with no organic slot. Normalize obvious variants (plurals, word order, stop words).
2. **Cluster by shared SERP, not string similarity.** Group queries Google answers with the same page: if two queries share roughly 4 or more of the same top-10 organic URLs, they belong in one cluster; below ~3 shared URLs, keep them separate. Never split a cluster only because wording differs; never merge two queries that surface different result types just because they share words.
3. **Separate clusters that demand different page formats.** When a cluster's SERP is dominated by a different result type (a guide vs. a comparison vs. a product page), split it - one page cannot satisfy two formats. Treat the coarse intent signal as a grouping input only; do not produce per-query intent labels or page-type recommendations here (that is search-intent-classifier).
4. **Build the pillar/cluster topology.** For each topic, define one broad pillar targeting the head term and the cluster pages targeting long-tail subtopics that report to it - a healthy pillar typically supports 5-15 cluster pages. Record the parent→child structure (which clusters belong under which pillar). Define structure only - leave link selection and anchor text to internal-linking-mapper.
5. **Pick the primary keyword per cluster.** Choose the one keyword the page can realistically win (highest relevant volume at achievable difficulty), then list the secondary keywords the same page should also satisfy. State difficulty honestly; on a low-authority domain, favor long-tail clusters with keyword difficulty under ~30 (on the standard 0-100 tool scales) first.
6. **Output a prioritized build order.** Score each cluster by opportunity = (revenue relevance × winnability) ÷ current coverage. Flag overlaps where two existing pages compete for one cluster (consolidate). Flag gaps where a cluster has demand but no page.
## Worked example
kinetic-typography
When your agent uses it
Put text in motion the right way - title cards, animated captions, lower-thirds, callouts, and word-by-word reveals - with a reading-time-per-word budget so copy holds long enough to read, plus enter/exit timing, weight and size transitions, and type hierarchy in motion. Use when someone asks to "animate this text", "make a title card", "animated captions/subtitles", "kinetic typography", "animate the headline", "lower-third", "word-by-word reveal", "text that pops in", or "how long should this line stay on screen". Do NOT use when the question is the cut, pacing, or easing curve of non-text motion - use motion-design-principles instead; do NOT use to plan which scenes or shots exist - use video-storyboard instead; do NOT use to resize or crop a caption track for a specific platform (9:16 TikTok, 1:1 feed) - use social-video-formatter instead; do NOT use for the color, contrast, or lighting of the type against its background - use motion-color-and-light instead.
What it loads into context
Frontmatter
241 tokensAlways available to the agent
SKILL.md
4,674 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Sibling skills - hand off, do not absorb
- Calculator
- Template: motion-type-spec
- references/reading-budget
- references/what-moves-what-holds
- references/handoff-to-remotion
Files
SKILL.mdcore18,373 bytesSKILL.mdcore18,373 bytesSKILL.mdcore18,373 bytes
Read SKILL.md
# Kinetic Typography
Text in motion has one job before it has any other: it must be readable. The most common failure in animated type is not ugliness, it is text that flashes past before the viewer finishes the first word, or text that animates so much the eye chases the motion instead of reading the words. This is the craft layer. It decides *how long every line stays, how it enters and leaves, and what is allowed to move* - and hands a concrete frame budget to `remotion-compose`, which writes the React/TypeScript that actually renders it.
The governing rule: **animate to serve the read, never to decorate it.** Motion earns its place only when it directs the eye, establishes hierarchy, or marks a beat. Everything below enforces that.
Work one running example throughout: a 16:9 product title card reading **"Ship video in one prompt"** (5 words), at 30fps, plus an animated lower-third caption track underneath.
## Workflow
Run these in order. Do not pick an entrance animation before you have the hold budget - the budget is what tells you whether you even have room to animate.
### Step 1 - Compute the reading-time-per-word budget first
kpi-scoreboard-and-cadence
When your agent uses it
Build a gym's KPI scoreboard, separate leading from lagging indicators, and set the daily/weekly/monthly meeting cadence that turns numbers into decisions. Use when a gym owner asks what KPIs to track, wants a scoreboard or numbers huddle, asks for a weekly meeting agenda, or wants to find the weakest stage of their lead-to-member funnel. Do NOT use to compute CAC, LTV, or the cash-flow money model - use gym-money-model instead; do NOT use to fix churn or the sales/booking conversion the scoreboard flags - use retention-and-churn-killer, objection-handling-and-speed-to-lead, or closer-sales-script instead.
What it loads into context
Frontmatter
146 tokensAlways available to the agent
SKILL.md
2,366 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Scoreboard template
- Meeting-agendas template
- KPI reference
- Cadence reference
Files
SKILL.mdcore9,621 bytesSKILL.mdcore9,621 bytesSKILL.mdcore9,621 bytesSKILL.mdcore9,621 bytes
Read SKILL.md
# KPI Scoreboard and Cadence
Pick the handful of numbers a gym runs on, organize them by funnel stage, and review them on a fixed rhythm where each meeting ends in a decision.
## Workflow
### Step 1: Track the KPIs by funnel stage
Organize metrics along the path from stranger to retained member, so a problem shows up at the stage it lives in. Definitions, formulas, sources, and healthy ranges are in the "KPI reference" section below.
- Leads: new prospects captured. Source: ad platform, forms, manual log.
- Booked: leads who scheduled a consult. Source: booking calendar.
- Shown: booked consults who attended. Source: consult log.
- Closed: shows who joined. Source: sales log.kubernetes-basics
When your agent uses it
Writes production-ready Kubernetes manifests - Deployments with correct resource requests/limits, readiness/liveness/startup probes, PodDisruptionBudgets, safe rollouts, and graceful shutdown. Use when someone asks "write a Deployment for my service", "my pods keep getting OOMKilled", "why do I get 502s during deploys", "what should my resource limits be", or "why did my pod restart in a loop". Do NOT use for local multi-container development environments - use docker-compose-wizard instead; for CI/CD pipelines that deploy to the cluster, use github-actions; for the Terraform that provisions the cluster itself, use terraform-expert.
What it loads into context
Frontmatter
153 tokensAlways available to the agent
SKILL.md
1,835 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Core objects
- Operating procedure
- Worked contrast
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytesSKILL.mdcore8,006 bytes
Read SKILL.md
# Kubernetes Basics
The gap between a manifest that runs and one that is production-ready is exactly the fields people omit: no resource requests means the scheduler packs pods blind and the node OOMs; no readiness probe means every rollout routes traffic to cold pods and users see 502s; no PodDisruptionBudget means a routine node drain takes the whole service down. This skill produces manifests where every one of those fields is set deliberately.
## Core objects
- Deployment: manages replica Pods and rolling updates.
- Service: stable network endpoint for a set of Pods.
[Credential-related source omitted]
- Ingress: HTTP routing into the cluster.
## Operating procedure
### Step 1: Gather inputslanding-page-copy
When your agent uses it
Write landing-page copy section by section - hero, social proof, features-as-benefits, objections, CTA - leading with one promise and proving it down the page. Use when someone asks "write my landing page", "improve my hero headline", "what should the homepage say", "turn these features into benefits", or is shipping a new page for a launch or campaign. Do NOT use for testing and optimizing an existing page's conversion rate - use landing-page-cro instead; for the one-line claim itself use positioning-statement; for the full message ladder use messaging-hierarchy.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
1,627 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Page skeleton template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,050 bytes
Read SKILL.md
# Landing Page Copy
A landing page makes one promise and spends the rest of the page making it believable. The costly mistake this skill prevents is the clever-but-vague page: paid traffic arrives, reads "Supercharge your workflow," and bounces before learning what the product does. Clarity outsells cleverness, and the reader decides in the first five seconds whether to give you thirty more.
## Operating procedure
Write top to bottom in the order below, because each section answers the objection the previous one created: the hero makes a claim, proof makes it credible, benefits make it concrete, objections clear the residue, and the CTA collects.
### Step 1: Gather inputs
1. The one promise - the single outcome the page sells. If the user offers three, make them pick one; the others become body copy. If no positioning exists yet, route to positioning-statement first.
2. The ICP and the pain in their own words (support tickets, sales calls, reviews - not the founder's paraphrase). Label paraphrases as guesses.
3. The real alternative (spreadsheet, incumbent, doing nothing).
4. Available proof: testimonials, logos, numbers, screenshots. Default: if none exists, use specificity of mechanism as interim proof and flag the gap.landing-page-cro
When your agent uses it
Improves an existing landing page's conversion rate through evidence-first CRO - audit heatmaps and recordings, write structured hypotheses, prioritize with ICE scoring, and run one-variable tests to statistical significance. Use when someone asks "why isn't my landing page converting", "audit this page before we scale spend", "which A/B test should we run first", or after a CVR drop or ahead of a paid-traffic push. Do NOT use for writing a new landing page from scratch - use landing-page-copy instead. Do NOT use for auditing the ad accounts sending the traffic - use paid-acquisition-audit instead. Do NOT use for reading the results of a finished experiment - use ab-test-analyzer instead.
What it loads into context
Frontmatter
160 tokensAlways available to the agent
SKILL.md
1,777 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: scored hypothesis backlog
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,244 bytesSKILL.mdcore8,244 bytes
Read SKILL.md
# Landing Page CRO
Conversion rate optimization is not applying best practices - it is forming specific hypotheses from real user evidence, testing them, and iterating. The costly failure this skill prevents is the redesign roulette: shipping a batch of untested "improvements" that mixes winners with losers, moves nothing, and teaches nothing, while paid spend keeps landing on a leaky page.
## Operating procedure
Order is strict: audit before hypotheses (a hypothesis without an observation is a guess), hypotheses before prioritization, prioritization before tests, and significance before verdicts.
### Step 1: Gather inputs
Collect before auditing. Label anything assumed as a guess.
1. Page URL and its single conversion goal (form fill, purchase, signup). A page with two goals is already a finding.
2. Current CVR and traffic: unique visitors per week and conversions per week - these decide the testing method in Step 5.language-learning
When your agent uses it
Coaches a language learner with a CEFR-targeted plan built on comprehensible input, spaced repetition, and early speaking practice, including daily minimums, a session template, and a weekly schedule. Use when someone says "help me learn Spanish", "I keep forgetting vocabulary", "how do I get from A2 to B1", or "design my daily language routine". Do NOT use for general exam or academic study techniques - use study-system instead.
What it loads into context
Frontmatter
107 tokensAlways available to the agent
SKILL.md
1,609 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: weekly schedule (A2 Spanish, 30 min/day, conversational B1 goal)
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,027 bytes
Read SKILL.md
# Language Learning
Languages are acquired through massive comprehensible input - material just above the learner's current level - plus consistent retrieval, not through grammar study alone. The costly failure this skill prevents is the intermediate plateau built from wrong-level habits: content so hard it burns motivation, flashcard decks that grow faster than they can be reviewed, and speaking postponed "until ready," which is a day that never arrives on its own.
## Inputs to collect
1. Target language. No default.
2. Current level on the CEFR scale (A1 beginner → C2 mastery). If unknown, estimate from a 2-minute probe - ask the learner to describe their day in the target language - and label the estimate a guess.
3. Goal: travel, conversation, work, or exam, plus the target CEFR level. Default: conversational B1.
4. Time budget per day. Default: 30 minutes. Below 20 minutes daily, reset expectations before planning (see thresholds).
5. Speaking access: tutor, exchange partner, or neither. Shapes Step 4.
## Operating procedure
language-reference
When your agent uses it
Use when a task needs dictionary or encyclopedia lookups from a live source - "define this word", "pronunciation / phonetics / synonyms of X", "get the Wikipedia summary of Y", "search Wikipedia for Z", or pulling a topic's intro paragraph and thumbnail into an app. Free Dictionary API covers words; Wikipedia's REST API covers topics; both keyless. Do NOT use for country facts or geography - use geo-places instead; do NOT use for trivia questions or random facts - use fun-content instead; if the request is a vague "I need live data", route through public-data-api-picker.
What it loads into context
Frontmatter
142 tokensAlways available to the agent
SKILL.md
1,253 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Ranked APIs
- Procedure
- URL templates + real response shapes
- Deliverable
- Do NOT use - reach for X instead
- Quality bar
Files
SKILL.mdcore5,345 bytesSKILL.mdcore5,345 bytes
Read SKILL.md
# Language & Reference Data
Look up words and topics against live keyless sources. The mistake this prevents: paraphrasing a definition or article summary from training memory when the task calls for a *citable* source - and, on the Wikipedia side, scraping HTML or hitting the heavyweight action API when the REST summary endpoint returns exactly the intro + thumbnail an app needs.
## Ranked APIs
1. **Free Dictionary API** (api.dictionaryapi.dev) - DEFAULT for word lookups: definitions, part of speech, phonetics with audio, synonyms/antonyms. English is reliable; other language codes exist but coverage is thin. No key.
2. **Wikipedia REST API** (en.wikipedia.org/api/rest_v1 and /w/rest.php/v1) - topic summaries, search, full page content. No key, but Wikimedia etiquette is enforced: send a real `User-Agent` (app name + contact), serialize requests, never hammer in parallel.
## Procedure
1. **Word vs topic.** A dictionary question ("define", "how do you pronounce", "synonym") → Free Dictionary. A knowledge question ("what is", "who was", "summary of") → Wikipedia.
2. **Wikipedia is two steps when the title is uncertain**: search first, then summary - guessing at title spelling gets 404s. Titles in URLs use underscores (`Albert_Einstein`) and are case-sensitive after the first letter.
3. **Attribute.** Wikipedia content is CC BY-SA - quote with a link to the page (`content_urls.desktop.page` in the summary response). Dictionary audio files carry their own license field.language-version-migrator
When your agent uses it
Ports a codebase across a breaking language or runtime version with compatibility shims, batched automated transforms, dual-runtime CI, and old-vs-new output diffing, ending in a flag-flip cutover with a rollback rule. Use when someone says "migrate us from Python 2 to 3", "we're jumping Node majors", "move to the next Java LTS", or any runtime upgrade where source must change to keep compiling or behaving correctly. Do NOT use for an application framework's major version (React, Rails, Spring) - use framework-upgrader instead; for proving a database schema or data migration is safe, use migration-safety-checker; and skip it entirely when the runtime change is purely operational (base image, CI matrix, deploy target) with no syntax or semantic breaks.
What it loads into context
Frontmatter
185 tokensAlways available to the agent
SKILL.md
1,832 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: migration plan skeleton
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore8,426 bytesSKILL.mdcore8,426 bytesSKILL.mdcore8,426 bytes
Read SKILL.md
# Language Version Migrator
Port code across a breaking language or runtime version by making it run on both versions first, then flipping the interpreter. The failure this skill prevents is the big-bang migration branch: a months-long fork that cannot merge, cannot ship, and cannot roll back, where the team discovers the semantic breaks (encoding, division, ordering) in production. Dual compatibility turns cutover into a config flip.
## Operating procedure
Run the steps in order. Dependencies gate the interpreter, tests gate the transforms, and the transforms gate the cutover - skipping ahead is how migrations stall.
### Step 1: Gather inputs
Collect these before planning. Label any estimate as a guess and refine as the inventory lands.
1. Source and target versions, and whether intermediate stops are required (Java 8 to 21 usually staged through 11 and 17; Python 2 to 3 is direct).
2. Codebase size: LOC and module count. This sets the batch count.
3. Test coverage per module. Any module under roughly 60 percent line coverage on migration-affected paths needs characterization tests first (pair with characterization-test-writer).launch-day-runbook
When your agent uses it
Use when you are running an actual launch day in real time across channels. Triggers on "launch day runbook", "we launch tomorrow", "Product Hunt launch day plan", "hour by hour launch schedule", "who does what on launch day", "PH / Hacker News / Twitter / email sequencing", "launch day checklist", "our launch is going sideways", "contingency if we get rate-limited / Show HN flops / site goes down". Turns the plan into a minute-by-minute war room. Do NOT use when you are still picking the launch date, beats, or channel mix - use [[launch-plan-sequencer]] for the calendar first; this is the execution companion that runs the day it lands on.
What it loads into context
Frontmatter
159 tokensAlways available to the agent
SKILL.md
3,175 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The ordered workflow
- Contingency table (decide these before, not during)
- Quality bar
- Do NOT
- Runnable artifact - launch-day timeline generator
- Deliverable
Files
SKILL.mdcore12,298 bytesSKILL.mdcore12,298 bytesSKILL.mdcore12,298 bytesSKILL.mdcore12,298 bytesSKILL.mdcore12,298 bytes
Read SKILL.md
# Launch Day Runbook
A launch is won or lost in the first six hours, and almost every failure is
operational, not strategic: the email went out before the Product Hunt page was
live, nobody was awake when the ranking slipped, the founder spent the morning
rewriting copy instead of replying to commenters. This skill is the war-room
runbook - an hour-by-hour script with named owners, a monitoring loop, and
pre-decided contingencies - so launch day is execution, not improvisation.
This is the *execution* companion. The strategy is already set upstream:
[[launch-plan-sequencer]] owns the date and beat calendar, [[positioning-statement]]
and [[messaging-hierarchy]] own the words, [[plg-motion-designer]] owns what
happens after signup, and [[sales-enablement-kit]] arms the team for the inbound
this creates. Do not re-litigate any of those today. Today you run the play.launch-email-sequence
When your agent uses it
Builds the open-cart email launch for a paid course - a 2-3 week pre-launch runway that seeds the problem, a day-by-day cart-open sequence of 6-8 emails over 5-7 days, honest deadline mechanics for the last-48-hours surge, and clicker/non-opener segmentation. Use when a creator says "write my launch emails", "plan my open cart sequence", "how many emails should I send during launch", or "how do I close the cart without fake scarcity". Do NOT use for always-on welcome or nurture automations - use email-drip-builder instead - or for the regular newsletter between launches - use email-newsletter-pro instead.
What it loads into context
Frontmatter
147 tokensAlways available to the agent
SKILL.md
2,022 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore8,765 bytes
Read SKILL.md
# Launch Email Sequence
An open-cart launch lives or dies on two things most creators get backward: the selling happens mostly before the cart opens, and 40 to 50 percent of sales land in the last 48 hours no matter how good the early emails are. Creators who skip the runway launch to a cold list, and creators who go quiet near the deadline - afraid of "too many emails" - leave half the revenue unsent. This skill builds both halves: the 2-3 week runway that seeds the problem, and the day-by-day cart sequence that serves the deadline instead of faking one.
This skill runs after course-outline-architect (the transformation statement is the campaign's spine) and in lockstep with course-sales-page - every email quotes that page's promise, value stack, and guarantee word-for-word, because a mismatch between email and page is where trust dies at the moment of purchase.
Worked example throughout: Dana, 8,000-subscriber list, launching "Notion OS for Consultants" at $297.
## Operating procedure
### Step 1: Collect the inputs
1. List size, recent open and click rates. Label estimates as estimates.
2. The transformation statement and module milestones from course-outline-architect.launch-plan-sequencer
When your agent uses it
Use when planning a product launch end-to-end and you need the full dated timeline. Triggers on "plan my launch", "launch plan", "how do I sequence a launch", "pre-launch checklist", "what happens before launch day", "launch timeline", "GTM launch plan", "coordinate a launch across channels", "who owns what for launch". Builds the pre-launch → launch-day → post-launch arc with a channel & asset checklist, owners, and dates. Do NOT use when you need the minute-by-minute run-of-show for the day itself - use launch-day-runbook instead. Do NOT use to write the core message - use positioning-statement and messaging-hierarchy. Do NOT use to design the self-serve adoption motion - use plg-motion-designer. Do NOT use to build the sales deck, battlecard, or demo script - use sales-enablement-kit.
What it loads into context
Frontmatter
201 tokensAlways available to the agent
SKILL.md
3,867 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The launch arc
- The channel & asset matrix
- Quality bar
- Do NOT
- Runnable artifact - the sequencer
- Deliverable
Files
SKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytesSKILL.mdcore14,200 bytes
Read SKILL.md
# Launch Plan Sequencer
A launch is not a day; it is an arc. The teams that botch launches treat "launch"
as a single deadline, pile every task onto it, and discover on the morning of that
the demo video is unrendered, sales has never seen the pitch, and nobody owns the
Hacker News thread. Your job is the opposite: stretch the launch into three phases -
**pre-launch**, **launch day**, **post-launch** - and assign every asset and
channel an owner and a date *before* the clock starts.
This skill owns the **sequence and the schedule**. It does not own the strategy
(what to launch, to whom, why now - that is a precondition the team brings in),
it does not own the message itself (that is [[positioning-statement]] and
[[messaging-hierarchy]]), and it does not own the minute-by-minute choreography
of the day itself (that is [[launch-day-runbook]]). You produce the map; thoselead-enrichment
When your agent uses it
Turns a raw or partial prospect list into a send-safe dataset by filling missing fields (work email, direct or mobile phone, title, firmographics, technographics) through a per-field provider waterfall, verifying every email, and stamping freshness. Use when someone says "enrich this list", "find the email for these people", "verify these emails before we send", "what do I do with catch-all domains", "this list is a year old, is it still good", "dedupe and clean this CSV", or "why is my bounce rate spiking". Do NOT use for selecting WHO to target and building the list itself - use prospect-list-builder. Do NOT use for tool-specific credit mechanics and filter syntax - use apollo-prospecting. Pair with cold-email-deliverability for the inbox-protection side of verification.
What it loads into context
Frontmatter
187 tokensAlways available to the agent
SKILL.md
3,262 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Enrichment plan template
- Worked example
- Triage calculator
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore13,440 bytesSKILL.mdcore13,440 bytesSKILL.mdcore13,440 bytes
Read SKILL.md
# Lead Enrichment
Enrichment is two jobs that people collapse into one and regret it: **filling** missing fields (a logistics problem - which provider has this contact, at what cost) and **verifying** the ones that touch deliverability (a risk problem - which addresses will bounce or hit a spam trap and torch the sending domain). A list is not "enriched" because the email column is full; it is enriched when every address intended for sending has been independently verified and stamped with a state, and every record carries a freshness date. The costly mistake this skill prevents: treating provider-supplied emails as send-ready. Vendors return pattern-generated guesses (first.last@domain) that were never SMTP-checked - send to those blind and expect a 12-20% bounce rate, which is how domains get throttled or blocklisted.
## Operating procedure
Follow in order: field definition gates the waterfall, the waterfall gates verification, and verification gates everything downstream. Skipping ahead means paying for fields you will not use or sending to addresses you never checked.
### Step 1: gather inputs
Collect before running anything. Label any guessed value as a guess.
- The list: row count, columns present, and source (CRM export, purchase, scrape, conference badge scan). Source predicts quality.
- The downstream play: email sequence, calls, LinkedIn, or multi-channel. This decides which fields are required.lesson-plan-builder
When your agent uses it
Builds a complete single-lesson plan - measurable objective, timed four-part arc (hook, direct instruction, guided practice, closure), a formative check with a proceed/re-teach rule, and differentiation notes - clear enough that a substitute could run it. Use when someone asks "write a lesson plan for...", "plan a 50-minute class on photosynthesis", "how should I structure tomorrow's lesson", or "turn this standard into a lesson". Do NOT use for sequencing a multi-week unit or full course - use curriculum-mapper instead; for deep per-learner adaptation of an existing lesson, use differentiated-instruction; for writing the quiz itself, use quiz-generator.
What it loads into context
Frontmatter
154 tokensAlways available to the agent
SKILL.md
1,558 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,020 bytes
Read SKILL.md
# Lesson Plan Builder
A strong lesson plan is a contract between the teacher and the learners: here is where we start, here is what we will do, here is how we will know we arrived. The costly failure this skill prevents is the lecture-heavy plan - the teacher talks for 30 minutes, practice gets squeezed into the last 5, and the first evidence of confusion arrives on the graded test, when it is too late to act.
## Operating procedure
Work the steps in order: the objective must exist before the exit task, and the exit task before the activities, because activities chosen first will smuggle in their own objectives.
### Step 1: Gather inputs
Collect before drafting. Where the user cannot answer, use the default and label the assumption a guess.
1. Subject and topic (required).
2. Grade or level (required - it sets vocabulary and pacing).life-coach
When your agent uses it
Runs structured coaching sessions using values clarification and the GROW model, ending every session with one committed action, a deadline, and an if-then plan for the likely obstacle. Use when someone says "I feel stuck in my life", "help me figure out what I want", "hold me accountable to my goals", or "coach me through this decision". Do NOT use for building a stress toolkit - use stress-management instead - or a journaling practice - use journal-framework; for a standing goal-tracking system, use goals-accountability. Coaching, not therapy: signs of clinical distress route to a licensed professional.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,547 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The coaching-therapy boundary (check first, every session)
- Inputs to collect
- Operating procedure
- Worked artifact: session close, bad vs. good
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,201 bytes
Read SKILL.md
# Life Coach
Coach by asking, not telling: the user owns the answers; the coach holds the structure and the mirror. The costly failure this skill prevents is the pleasant, drifting conversation that ends in insight but no commitment - sessions that feel good and change nothing. Every session ends with exactly one action, a deadline, and a plan for the obstacle most likely to stop it.
## The coaching-therapy boundary (check first, every session)
Coaching works with a functioning present toward a chosen future. It is not therapy, and the line is not negotiable. If any of these surface - hopelessness or worthlessness that persists across sessions, any mention of self-harm or suicide, panic that interferes with daily functioning, trauma processing, substance dependence, an eating disorder, or grief that is not moving - stop coaching on that thread. Say plainly: this deserves more than coaching can offer, and a licensed therapist or physician is the right professional. Provide encouragement to seek one, and continue coaching only on clearly separate, forward-looking topics if the user wants to. Never attempt techniques from therapy (trauma exploration, diagnosis, treatment). Routing someone to real help is a coaching success, not a failure.
## Inputs to collect
1. The topic the user wants coaching on, in their own words.
2. Time horizon: a single decision, a 90-day push, or a life-direction question. Default: 90 days.
3. Prior attempts: what they have tried and what happened - the obstacle usually lives in this answer.
4. Accountability preference: check-in cadence and channel. Default: one check-in per week.lifecycle-journey-map
When your agent uses it
Builds a customer lifecycle journey map with behavioral entry/exit criteria per stage, one goal and one signal per stage, message briefs, and dead-zone flags. Use when someone asks "map our customer lifecycle", "where are the gaps in our lifecycle emails", "what should we send at each stage", or is aligning a team on lifecycle messaging strategy. Do NOT use to write the actual email sequence - use email-drip-builder for onboarding drips, win-back-campaign for dormant users, or push-notification-copy for push messages; this skill produces the map those campaigns are built on.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,428 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked example - B2B invoicing SaaS
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,696 bytesSKILL.mdcore6,696 bytes
Read SKILL.md
# Lifecycle Journey Map
A lifecycle map answers one question for every moment a user can be in: what should they feel, know, or do next? Get this right and every downstream campaign has a clear job; skip it and the team ships overlapping messages with no owner, no goal, and no way to know if they worked. The costly mistake this skill prevents is calendar-based messaging - sending on day 7 regardless of what the user actually did.
## Inputs to collect
1. **Product and its first value moment** - the action that proves a user got value (first export, first invoice sent, first report shared). If unknown, propose one and label it a guess to validate against activation data.
2. **Tracked events** - the map can only trigger on events the product already tracks. Get the event list before designing transitions.
3. **Drop-off data per stage transition** - signup→activation rate, activation→habit rate, renewal rate. If unavailable, mark the map "unvalidated" and make pulling this data step 6.
4. **Existing touchpoints** - every email, push, and in-app message currently sent, with its trigger.
5. **Channels available** - email, in-app, push, SMS.
## Operating procedure
linear-workflow
When your agent uses it
Configures a Linear team end to end - team structure, minimal workflow states, cycles, a small label taxonomy, a daily triage rotation, and priority SLAs with breach views. Use when someone asks "set up Linear for my team", "how should we structure teams and labels in Linear", "our Linear backlog is a mess", or "how do we handle inbound bugs in Linear". Do NOT use for writing the tickets themselves - acceptance criteria, story points, Definition of Ready - use jira-ticket-writer instead (its ticket anatomy applies to Linear issues too). For deciding sprint scope and capacity, use sprint-planning.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,751 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: team setup spec
- Deliverable
- Do NOT
- Quality bar
- Related skills
Files
SKILL.mdcore7,427 bytes
Read SKILL.md
# Linear Workflow
Linear rewards lightweight, opinionated process; teams that port over a heavyweight Jira scheme get the worst of both worlds - ceremony without clarity. This skill sets up a team so issues flow predictably from idea to done, and inbound noise never silts up the backlog.
## Inputs to collect
1. Delivery units: which groups actually ship together? (Squads, not departments.)
2. Team size and count - decides whether sub-teams are justified yet.
3. Inbound volume: roughly how many bugs/requests arrive per week, and from where.
4. Current cadence, if any (1- or 2-week sprints, or none). If unknown, default to 2-week cycles and label it a default.
5. The 3-5 product areas people already use in conversation ("auth", "billing") - these become area labels.
## Operating procedure
linkedin-post-writer
When your agent uses it
Write LinkedIn posts built for the feed - a hook that survives the two-line truncation, scannable one-idea-per-line structure, a comment-bait close, and first-hour engagement moves. Use when someone asks "write a LinkedIn post about this", "punch up my LinkedIn hook", "why did my post get no reach", or "turn this win into a LinkedIn post". Do NOT use for Twitter/X threads - use tweet-thread-builder instead. Do NOT use for scheduling across a monthly calendar - use social-content-calendar instead.
What it loads into context
Frontmatter
133 tokensAlways available to the agent
SKILL.md
1,590 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: bad vs good
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,937 bytes
Read SKILL.md
# LinkedIn Post Writer
Write posts for how people actually read LinkedIn: on a phone, fast, in a crowded feed. The feed truncates after roughly two lines, so the costly mistake is spending effort on the body while line one fails to earn the "see more" click - and then abandoning the post at publish, when the first hour of engagement decides whether the algorithm shows it to anyone at all.
## Operating procedure
### Step 1: Gather inputs
1. The story or idea - one concrete moment, result, or claim. If the user offers a vague topic ("leadership"), push for a specific incident or number; specificity is the post.
2. The single takeaway a reader should leave with (one, not five).
3. Proof details: numbers, timeframes, names of situations (anonymized as needed). Label estimates as estimates.
4. The engagement goal: comments (default - comments drive reach hardest), profile visits, or a lead-magnet offer.
5. Whether the user can be online for 60 minutes after posting (this changes the close - see Step 5).
listing-description-writer
When your agent uses it
Writes MLS and portal listing copy that sells the property - lead with the differentiator, translate features into benefits, specifics over adjectives, platform length limits, and a fair-housing-safe banned-claims list. Use when an agent says "write my listing description", "make this MLS copy better", "describe this house for Zillow", or "punch up my listing remarks". Do NOT use for e-commerce product listings - use product-description-writer instead, or amazon-listing-optimizer for Amazon listings.
What it loads into context
Frontmatter
110 tokensAlways available to the agent
SKILL.md
1,973 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Platform length limits
- Feature → benefit translation, worked
- Banned-claims list (hard gate)
- Good/bad pair - same property, 742 Alder Court
- Deliverable
- Do NOT
- Quality bar
- Escalation and compliance
Files
SKILL.mdcore8,955 bytesSKILL.mdcore8,955 bytesSKILL.mdcore8,955 bytesSKILL.mdcore8,955 bytes
Read SKILL.md
# Listing Description Writer
A listing description has one job: make the right buyer book a showing. Most listing copy fails by burying the one thing that makes the house different under a pile of interchangeable adjectives - "stunning", "charming", "must-see" - that buyers have learned to skim past. The costly mistake this skill prevents is worse than boring copy: a fair-housing violation or an inaccurate claim that survives into a lawsuit. Describe the property and the amenities, never the people who might live there.
Worked example throughout: Priya, a residential agent in a suburban market, 14 transactions last year at a $485,000 average sale price, building toward 24. Her current listing is 742 Alder Court - a 4-bed, 2.5-bath, 2,340 sq ft colonial built in 1998 on a fenced 0.28-acre lot, with a 2022 kitchen renovation (quartz waterfall island), a new roof, and a location 0.4 miles from the commuter rail station. It lists at $499,000 - the price story behind that number lives in cma-narrative-builder, which shares this same worked listing.
## Operating procedure
1. **Collect the facts and verify them** (inputs below). Every claim in the copy must trace to a verifiable fact - a permit, a receipt, the listing agreement, or the agent's own inspection. Unverified claims come out or get softened before drafting starts, because retrofitting accuracy into finished copy always misses something.
2. **Pick the single differentiator.** One thing this property has that the competing inventory at the same price band does not. Ask: "What did the seller pay for, or what does the location give, that the buyer cannot get next door?" For 742 Alder Court it is the renovated kitchen plus rail proximity, not the bed count - every competitor at $499k has four bedrooms.
3. **Lead with the differentiator.** The first sentence carries the differentiator with a specific detail. Never open with beds/baths/sq ft - the portal already displays those above the copy.
4. **Translate the top five features into benefits.** Feature → what it lets the buyer do. "New roof" → "no five-figure surprise in year one." Run the translation table below.
5. **Apply the specifics-over-adjectives pass.** Replace every subjective adjective with a nameable noun. "Gorgeous kitchen" becomes "quartz waterfall island and a 36-inch gas range." If a sentence would survive being pasted into any other listing, it is too generic - rewrite it.
6. **Run the banned-claims check** (list below). This is a hard gate, not a suggestion.literature-review
When your agent uses it
Produces a thematic literature review that organizes a field into an argument - consensus, disputes, evidence quality, and the open gap the reader should care about. Use when someone asks "write my related-work section", "review the literature on X", "what does the research say about X", or "organize these papers into themes". Do NOT use for an exhaustive, protocol-registered review with formal inclusion criteria and risk-of-bias appraisal - use systematic-review instead. Do NOT use for synthesizing interviews or mixed non-academic sources into decision themes - use research-synthesis instead.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
1,217 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore5,955 bytes
Read SKILL.md
# Literature Review
A literature review is an argument about the state of knowledge, not a list of summaries. The failure it prevents is the annotated bibliography in disguise - one paragraph per paper, no comparison, no position - which readers skip and reviewers reject because it demonstrates reading, not understanding. Every section must compare works against each other and end somewhere: a consensus, a live dispute, or a gap.
## Operating procedure
### Step 1: Gather inputs
Collect before writing:
- The guiding question or claim the review serves (a thesis chapter, a paper's related-work section, a research proposal, a decision memo). Default framing if none given: "What is known, what is contested, and what is missing about X?"
- The corpus: papers supplied, or search scope (fields, venues, timeframe) if collection is part of the job.
- Audience and length target (a related-work section runs 800-1,500 words; a standalone review 3,000+).
- Inclusion posture: representative coverage is the default; if the user says "every study, defensibly complete," stop and route to systematic-review - that job needs a protocol.llm-evaluation
When your agent uses it
Builds evaluation harnesses for LLM products - golden datasets, deterministic checks, calibrated LLM-as-judge rubrics, and CI regression gates that turn "seems good" into a tracked number. Use when someone asks "how do I know this prompt change didn't break anything", "set up evals for my RAG pipeline", "is LLM-as-judge reliable", "why did quality drop after the model swap", or is shipping an LLM feature with no quality measurement. Do NOT use for classical ML model reporting (precision/recall, ROC curves, confusion matrices on trained classifiers) - use model-evaluation-report instead; do NOT use for analyzing online A/B experiments - use ab-test-analyzer instead.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,788 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: calibrated judge rubric
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,057 bytes
Read SKILL.md
# LLM Evaluation
A prompt or model change that looks fine on the three examples you tried will break inputs you did not try, and without an eval you find out from users. The costly mistake this skill prevents is shipping on vibes: an uncalibrated judge or a ten-example "test" gives you a number that moves randomly, so real regressions hide inside noise. Build the dataset first, calibrate the judge before trusting it, and gate changes in CI.
## Operating procedure
Follow the order: dataset before metrics, calibration before judging, pipeline before iteration. A judge calibrated against nothing is a random-number generator with confidence.
### Step 1: Gather inputs
Collect these before building anything. Where the user cannot answer, use the default and label the value a guess.
1. Task definition - what the system produces, in one sentence.
2. What "good" means - 3-5 quality dimensions (correctness, groundedness, tone, format). Default: correctness plus format validity.load-testing
When your agent uses it
Designs and interprets load tests - smoke, load, stress, soak, and spike - with realistic ramp profiles, percentile-based pass/fail thresholds, and a bottleneck-reading procedure, producing a runnable test plan. Use when someone asks "how many users can we handle", "how do I write a k6 or Gatling test", "will this survive the launch", "why did the site die at 500 users", or is preparing for a traffic event. Do NOT use for diagnosing frontend page-speed or Core Web Vitals - use web-performance instead - and do NOT use for tuning database connection pools found as the bottleneck - use connection-pool-tuner instead.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
2,143 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: test plan skeleton
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,343 bytesSKILL.mdcore8,343 bytes
Read SKILL.md
# Load Testing
The point of a load test is to find the capacity limit and the failure mode before users do, and to turn "will it hold?" into a number with a date on it. The costly mistake this skill prevents is the vanity test: hammering one cached endpoint from one laptop, reading the average latency, and declaring readiness - a test that passes while production dies at one-third the "tested" load.
## Operating procedure
### Step 1: Gather inputs
Collect before writing any script; label estimates as guesses.
1. Target load: expected peak in requests/sec or concurrent users, and where the number came from (analytics, marketing forecast, guess).
2. SLOs: the p95/p99 latency and error-rate targets that define pass/fail. If none exist, propose p95 < 500ms, p99 < 1500ms, error rate < 1% as defaults and get sign-off.
3. Top 3-5 user journeys with their real traffic mix (e.g. 80% browse, 15% search, 5% checkout) - from production access logs, not intuition.
4. Environment: production-like staging (same instance sizes, same data volume) or coordinated production test. Toy data lies: a 1,000-row table performs nothing like a 100M-row one.local-referral-engine
When your agent uses it
Builds referral flow and local partnerships for a local service business - the ask-after-five-star-moment rule, two-sided referral incentive design, complementary-trade partnerships, and neighborhood-domination tactics like job-site signage and just-worked-on-your-street cards. Use when an owner asks "how do I get more word-of-mouth", "should I pay customers for referrals", "how do I partner with other trades", or "how do I get more jobs in the same neighborhood". Do NOT use for gym referral and affiliate programs - use referral-and-affiliate-system instead. Do NOT use for formal B2B partnership deals - use partnership-strategy instead.
What it loads into context
Frontmatter
140 tokensAlways available to the agent
SKILL.md
2,181 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to elicit
- Concrete thresholds
- Worked artifact: Summit's referral math
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore9,667 bytesSKILL.mdcore9,667 bytesSKILL.mdcore9,667 bytesSKILL.mdcore9,667 bytes
Read SKILL.md
# Local Referral Engine
Referrals are the best customers a local service business gets: near-zero
acquisition cost, pre-sold trust, and higher close rates than any paid
channel. The costly mistake this skill prevents is leaving referrals passive -
"we get word of mouth" - instead of building the three systems that
manufacture them: a timed ask, a two-sided incentive, and a partner network.
Run local-unit-economics first to see the prize: Summit Plumbing (4 techs,
$450 average ticket, $75,000 a month) pays about $119 to acquire a paid-lead
customer, so every referred customer is roughly $119 of margin that paid
channels never touch.
## Operating procedure
local-seo-playbook
When your agent uses it
Ranks a local service business in the Google map pack and local organic results by working the three local ranking factors - relevance, distance, and prominence - through service-page structure, honest city pages, citation consistency, and review velocity. Use when an owner asks "how do I rank in the map pack", "why does my competitor outrank me locally", "do I need a page for every service", or "should I make pages for every city near me". Do NOT use for hands-on Google Business Profile settings - use google-business-profile-optimizer instead. Do NOT use for general non-local website SEO - use seo-optimizer instead.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
2,082 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The three factors and what actually moves each
- Operating procedure
- Inputs to elicit
- Concrete thresholds
- Worked artifact: Summit Plumbing site structure
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore9,411 bytes
Read SKILL.md
# Local SEO Playbook
Local ranking is not general SEO with a city name bolted on: Google ranks the
map pack on three factors - relevance, distance, and prominence - and only
two of them can be moved. The costly mistake this skill prevents is spending
months on tactics that move none of the three (blogging into the void,
doorway city pages) while the top competitor compounds reviews and citations.
Before starting, run local-unit-economics: for Summit Plumbing (4 techs, $450
average ticket, $75,000 a month, roughly $119 CAC on paid leads), every
organic customer displaces a paid one, which is the budget case for the hours
this playbook takes.
## The three factors and what actually moves each
local-unit-economics
When your agent uses it
Models the money engine of a local service business - cost per lead by channel, booking rate, show rate, close rate, average ticket, and repeat rate - computing CAC, first-90-day cash, LTV to CAC, and a self-funding verdict, with a runnable calculator. Use when an owner asks "what does a new customer actually cost me", "can I afford more ad spend", "is my marketing paying for itself", or "which lead channel should get the next dollar". Do NOT use for gyms - use gym-money-model instead. Do NOT use for SaaS or subscription software - use unit-economics instead.
What it loads into context
Frontmatter
139 tokensAlways available to the agent
SKILL.md
2,919 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to elicit
- Calculator
- Template: channel-economics-sheet
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore11,587 bytesSKILL.mdcore11,587 bytesSKILL.mdcore11,587 bytes
Read SKILL.md
# Local Unit Economics
Install this skill first. Every other local-service growth skill -
google-business-profile-optimizer, review-engine, local-seo-playbook,
quote-follow-up-sequences, no-show-reduction, local-referral-engine - spends
money or staff time, and this model is what says whether that spend pays for
itself. The costly mistake it prevents is judging marketing by cost per lead
instead of cost per acquired customer: a cheap lead that never books, shows, or
closes is the most expensive lead there is.
Work the example throughout: Summit Plumbing, a 4-tech residential plumbing
company in a mid-size metro, average job $450, doing $75,000 a month - about
167 jobs.
localization-guide
When your agent uses it
Adapts English content for a target language and region - swapping idioms, setting the right formality register, replacing cultural references, converting dates, units, and currency - and delivers the localized text with a notes column explaining each adaptation and flagging what needs a native reviewer. Use when someone asks "localize this copy for the Japanese market", "adapt our marketing copy for the Mexican market", "make this translation sound native instead of literal", or a literal translation reads foreign, stiff, or risky. Do NOT use for auditing in-product UI text quality - use ux-writing-audit instead.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
787 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Translation vs. localization
- Process
- What needs adaptation
- Tone and voice
- Sensitivity
- Process discipline
- Honesty and limits
- Anti-patterns
- Output
Files
SKILL.mdcore3,817 bytesSKILL.mdcore3,817 bytes
Read SKILL.md
# Localization Writer
You adapt content for other languages and cultures. Localization is more than translation: it's making content feel native - appropriate in idiom, formality, references, and tone for the target market.
## Translation vs. localization
- Translation converts words. Localization adapts meaning, tone, and cultural fit.
- A literally correct translation can still feel foreign, awkward, or even offensive. You catch and fix that.
## Process
1. Gather: the source content, the target locale (language AND region - es-MX differs from es-ES), the content type, and the brand voice.
2. Identify everything that won't transfer literally (below).
3. Adapt, flag, and explain choices for a human reviewer.logo-brief-writer
When your agent uses it
Writes a complete creative brief for a logo or identity project - business context, a three-adjective personality axis, deliverables and constraints, competitive territory, rounds and timeline, and concrete success criteria. Use when someone asks "write a logo brief", "brief a designer for our rebrand", "what do I send the identity studio", or is kicking off a brand identity engagement. Do NOT use for generating the brand name itself - use brand-naming instead; for documenting the finished identity system's usage rules, use brand-guidelines; for briefing broader creative direction, use moodboard-builder.
What it loads into context
Frontmatter
138 tokensAlways available to the agent
SKILL.md
1,600 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template: logo brief
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,237 bytesSKILL.mdcore7,237 bytesSKILL.mdcore7,237 bytes
Read SKILL.md
# Logo Brief Writer
A strong creative brief is the contract between strategy and design: it gives the designer everything they need and nothing they don't. The costly failure is the vague brief - "professional but approachable, we'll know it when we see it" - which produces vague concepts, three rounds of subjective feedback, and a dispute about what was in scope. This procedure forces the decisions that vagueness defers.
## Operating procedure
### Step 1: Gather inputs
Extract these before writing a word. If the client cannot answer, that gap goes in the brief as an open question - a brief written on silently-guessed inputs produces vague logos.
- What the brand does, who its primary customer is, what problem it solves, and what makes it distinct from the nearest competitor.
- Project stage: new brand, rebrand, or refresh - and the reason for the project now. This changes everything about creative direction (a refresh must preserve equity; a rebrand must explain what is being escaped).
- Any locked constraints: existing assets to retain, color restrictions, trademark concerns, naming already fixed (if naming is open, route to brand-naming first - a logo cannot be briefed against an unsettled name).
- Budget band and hard deadline, if known. Default: unknown, flagged as open.longevity-protocol
When your agent uses it
Assembles an evidence-tiered weekly longevity routine across the four pillars with the strongest healthspan data - sleep, zone 2 aerobic work, strength training, and dietary pattern - plus stress and connection, delivered as a filled weekly template with honest evidence labels. Use when someone asks "build me a longevity routine", "what actually extends healthspan", "audit my health habits", or "is this supplement stack worth it". General wellness guidance, not medical advice. Do NOT use to program the individual pieces in depth - use zone-2-cardio-plan for the aerobic block, strength-training-plan for lifting, sleep-optimizer for sleep, nutrition-planner for macros; for HIIT/conditioning use fitness-program; for interpreting labs use bloodwork-explainer.
What it loads into context
Frontmatter
171 tokensAlways available to the agent
SKILL.md
1,907 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Evidence tiers - apply to everything
- Operating procedure
- Worked artifact
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,851 bytes
Read SKILL.md
# Longevity Protocol
The costly mistake in longevity is inversion: obsessing over supplements and biohacks (weak evidence) while sleep, aerobic capacity, strength, and diet (overwhelming evidence) go under-managed. This skill assembles a weekly routine strictly in order of evidence strength and labels every recommendation with its tier, so effort lands where the healthspan return actually is. Healthspan - years in good physical and cognitive function - is the target, not lifespan alone.
## Evidence tiers - apply to everything
- **Tier 1 (act on these):** sleep, exercise (aerobic + strength), dietary pattern, not smoking, social connection. Large, consistent human outcome data.
- **Tier 2 (reasonable, weaker data):** time-restricted eating, sauna, specific dietary tweaks beyond the core pattern. Plausible mechanisms, mixed or smaller human trials.
- **Tier 3 (honest label: mostly unproven for longevity):** nearly all supplements marketed for longevity - NAD precursors, resveratrol, and similar. Mechanistic or animal data, little or no human outcome evidence. Never present a Tier 3 item as if it were Tier 1, and never let it displace Tier 1 effort or budget.
## Operating procedure
Order matters: pillars are addressed in descending evidence-and-leverage order, and the weekly template is assembled only after every pillar has a concrete dose.
m-and-a-analysis
When your agent uses it
Structures an acquisition case - one-line thesis, strategic-fit test, risk-haircut synergy bridge with cost synergies cut 20-30% and revenue synergies cut 50%+, valuation with a written walk-away price, and integration risk - into a screening scorecard and deal memo. Use when someone asks "should we buy this company", "evaluate this acquisition target", "are these synergy numbers real", "what would we pay for them", or is preparing a board memo on a deal. Do NOT use for building the diligence request list and workstream tracker once a deal is moving - use due-diligence-checklist instead; for assessing whether the combined business has a durable advantage, pair with competitive-moat.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,789 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: screening scorecard
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,149 bytes
Read SKILL.md
# M&A Analysis
Most acquisitions destroy value, and they fail not on the spreadsheet but on the thesis and the integration. The costly failure this skill prevents is the deal that "works" only because unhaircut synergies were stacked until the price looked justified. Judge the deal on strategy first and synergies second, and write the walk-away price before anyone falls in love.
## Inputs to collect
1. **The target**: what it does, revenue and growth, headcount, ownership.
2. **The acquirer's strategy**: the plan the deal is supposed to serve - in writing, not implied.
3. **Price expectations**: asking price or range, last financing valuation, comparable transactions if known.
4. **Claimed synergies**, itemized and separated into cost and revenue, with whoever claimed them named.
5. **Key people and customers**: founder/team dependence, top-customer concentration.
6. If any input is a guess, label it a guess and carry it through the scorecard as such.
## Operating proceduremanuscript-reviser
When your agent uses it
Use when revising fiction prose that ALREADY EXISTS - a second/third-draft pass on a chapter or scene, a continuity or POV-consistency check, fixing a scene that drags, cutting dead weight, or tightening overwritten prose at the line level. Do NOT use to draft a brand-new scene from nothing (use Fiction Scene Writer), to build or mimic an author's voice (use Ghostwriter), or for neutral encyclopedic prose (use Wikipedia Style Writer); to outline or re-architect the plot at the beat level, use Story Structure Architect.
What it loads into context
Frontmatter
131 tokensAlways available to the agent
SKILL.md
1,217 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Before you touch a line
- The revision passes - run in this order
- The quality bar - a revision is done well when
- How to deliver
- What NOT to do
Files
SKILL.mdcore5,599 bytes
Read SKILL.md
# Manuscript Reviser
You revise existing fiction. The draft is already written; your job is to make it sharper, tighter, and internally consistent without rewriting it into your own voice. Revise, don't replace.
## Before you touch a line
1. Read the whole passage once, all the way through, before changing anything. You cannot judge pacing or continuity from a fragment.
2. Establish the ground truth you must protect: the POV character and the narrative distance (close third, omniscient, first), the verb tense, and the author's existing voice. These are constraints, not targets - preserve them unless the user asked you to change them.
3. Ask for the surrounding context if a continuity or POV call depends on material you can't see (an earlier chapter, a character sheet, established timeline). Don't guess at canon.
## The revision passes - run in this order
Work coarse-to-fine. Fixing structure after polishing sentences wastes the polish.
market-sizing
When your agent uses it
Builds a defensible TAM/SAM/SOM market-sizing model with bottom-up and top-down triangulation and a low/base/high sensitivity range. Use when someone asks "how big is the market", "calculate my TAM", "size this opportunity", is preparing an investor deck market slide, or needs to defend a market number under diligence. Do NOT use for revenue projections of an existing business - use revenue-modeling instead.
What it loads into context
Frontmatter
101 tokensAlways available to the agent
SKILL.md
741 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The Three Layers
- Build Bottom-Up First
- Triangulate Top-Down
- SAM and SOM
- Show the Math
- Common Mistakes
- When the Market Is New
- Deliverable
Files
SKILL.mdcore3,267 bytes
Read SKILL.md
# Market Sizing
A market size number is only useful if it is defensible. Investors have seen a
thousand "100B TAM" slides and discount all of them. This skill builds a number
you can defend under questioning by showing your work.
## The Three Layers
- **TAM (Total Addressable Market)**: total spend if everyone who could buy,
did, at your price. The ceiling.
- **SAM (Serviceable Addressable Market)**: the slice you can actually reach
given product, geography, and segment.
- **SOM (Serviceable Obtainable Market)**: what you can realistically win in
3-5 years given competition and execution. The number that matters formarketing-attribution
When your agent uses it
Guides selection and honest application of a marketing attribution model, surfacing common measurement traps. Use when evaluating channel ROI, auditing tracking, or choosing between attribution models.
What it loads into context
Frontmatter
49 tokensAlways available to the agent
SKILL.md
1,113 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Understand What Each Model Assumes
- Choose Based on Sales Cycle and Data Volume
- Avoid the Three Most Common Traps
- Triangulate, Do Not Depend on One Signal
- Worked Example: Same Data, Two Models, Two Decisions
- Document and Communicate the Model
- Deliverable
- Quality bar
Files
SKILL.mdcore5,711 bytes
Read SKILL.md
# Marketing Attribution
Attribution does not reveal truth - it reveals a model's assumptions about truth. Choosing a model is choosing a set of tradeoffs. The goal is to pick the model that creates the least-distorted decisions for the business, not the one that flatters the most channels.
## Understand What Each Model Assumes
First-touch attribution credits the first interaction entirely - useful for understanding discovery channels, but it ignores everything that closed the sale. Last-touch does the opposite and over-credits closers like brand search. Linear attribution spreads credit evenly - neutral but often meaningless. Time-decay weights recent touches more heavily - reasonable for short sales cycles. Data-driven attribution uses conversion path modeling - best in theory but requires high conversion volume (typically 1,000+ monthly conversions) and complete tracking. Position-based (U-shaped) splits 40/20/40 between first touch, middle, and last touch - a practical default for teams that lack volume for data-driven models.
## Choose Based on Sales Cycle and Data Volume
For sales cycles under 7 days with fewer than 500 monthly conversions: use last-touch with branded/direct excluded. For cycles 7-30 days: use position-based. For cycles over 30 days or complex B2B: use time-decay or a CRM-based opportunity model. For teams with 1,000+ conversions and clean tracking: test data-driven and compare decisions it produces versus current model. Never use first-touch as a primary model for budget allocation - it systematically over-invests in awareness and starves conversion channels.
## Avoid the Three Most Common Traps
mcp-builder
When your agent uses it
Design and build Model Context Protocol servers that models use correctly - choosing stdio vs stateless Streamable HTTP while still serving 2025-era clients, shaping a few task-level tools with clear descriptions, JSON Schema inputs and honest annotations, bounding and paginating output, returning recoverable errors as isError results, securing remote servers with OAuth 2.1, and testing with the MCP Inspector. Use when someone asks "build an MCP server for our API", "turn this service into MCP tools", "why does the model keep calling the wrong tool", "add OAuth to my remote MCP server", or "my stdio server keeps disconnecting". Do NOT use for designing a REST or GraphQL API consumed by human-written code - use api-design instead - or for coordinating multiple LLM agents - use agent-orchestration instead.
What it loads into context
Frontmatter
186 tokensAlways available to the agent
SKILL.md
2,309 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example - TypeScript SDK v2, stdio
- Thresholds
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore9,999 bytes
Read SKILL.md
# MCP Server Builder
An MCP server is an API whose caller is a language model. It picks tools by name and description, pays in context for every byte returned, and cannot read your docs. The costly mistake this skill prevents is wrapping a REST API one endpoint per tool. That produces forty thin tools, vague descriptions, unbounded JSON and errors thrown as protocol failures, and the result is a server the model misuses, that floods the context window, and that fails silently.
## Operating procedure
### Step 1: Write the task list before the tool list
List the 5-10 concrete things users will ask for, such as "find open bugs assigned to me". Each task should take one or two calls. One `search_issues` with filters beats three list-and-filter tools; chain upstream calls inside the server.
### Step 2: Choose the transport and protocol posture
- **stdio** is for a local subprocess (filesystem, git, CLIs). On stdio, stdout belongs to the protocol: log to stderr, never `console.log` or `print`.
- **Streamable HTTP** is for remote or shared servers. Since spec revision 2026-07-28 the protocol is stateless:meal-planner
When your agent uses it
Builds a realistic weekly meal plan - protein anchors first, a batch-cook prep schedule, and a deduplicated grocery list grouped by store section - sized to household, budget, and weeknight time limits. Use when someone asks "plan my meals for the week", "make me a grocery list", "what should we eat this week", or "help me meal prep on Sunday". Do NOT use for setting calorie and macro targets or nutrition strategy - use nutrition-planner instead; this skill turns targets into cookable weeks.
What it loads into context
Frontmatter
123 tokensAlways available to the agent
SKILL.md
1,670 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: weekly plan template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,282 bytes
Read SKILL.md
# Meal Planner
Plan meals people will actually cook: realistic effort, minimal waste, and a grocery trip that happens once. The costly failure this skill prevents is the aspirational plan - seven distinct ambitious dinners that collapse by Wednesday into takeout, with a fridge of unused specialty ingredients rotting as a reminder. Repetition and batch-cooking are features, not compromises.
This skill owns planning logistics: what to cook, when to prep it, and what to buy. Nutrition strategy - calorie targets, macro splits, protein per kilogram - belongs to nutrition-planner. When the user brings targets from there, honor them per meal; when they bring none, default each main meal to protein + vegetable + starchy carb and do not invent numeric targets.
## Inputs to collect
Ask for what's missing; label assumptions as guesses.
1. People and appetites (adults, kids). Default: 2 adults.
2. Days and meals to cover. Default: 5 weeknight dinners plus lunches from leftovers.
3. Restrictions and allergies. Allergies are absolute - a plan that violates one is a failed plan, not a flawed one.
4. Calorie/macro targets, if any (typically from nutrition-planner). Default: none; balanced plate rule applies.media-monitor
When your agent uses it
Turns raw brand mentions across news and social into an actionable comms picture - precise monitoring queries, grounded sentiment classification, numeric alert thresholds for narrative shifts, severity triage, and a daily digest a comms team can act on within minutes. Use when someone asks "set up brand monitoring", "are people talking about us and how", "is this negative story spreading", "design our mention alert thresholds", or "write our daily media digest". Do NOT use for spotting broader market or cultural trends beyond a tracked brand - use trend-analysis instead - or for gathering intelligence on competitors' moves - use competitive-intelligence instead.
What it loads into context
Frontmatter
140 tokensAlways available to the agent
SKILL.md
1,576 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: daily digest template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,076 bytesSKILL.mdcore7,076 bytesSKILL.mdcore7,076 bytesSKILL.mdcore7,076 bytesSKILL.mdcore7,076 bytes
Read SKILL.md
# Media Monitor
Raw mention feeds are noise; the costly failures are symmetric - missing the forum post that becomes a national story eighteen hours later, and escalating a comms fire drill over three sarcastic tweets. This skill turns mentions into an actionable picture: queries precise enough to trust, sentiment grounded in text, numeric thresholds that separate shift from chatter, and a digest that lets a comms team act within minutes.
## Operating procedure
Query design comes first because every downstream number - volume, sentiment, spike detection - inherits the query's precision; a noisy query makes every alert a false one.
### Step 1: Gather inputs
1. Entities to track: brand names, products, executives, competitors, campaign hashtags.
2. Known ambiguities: homonyms, generic words in the name, celebrity name collisions.
3. Sources in scope and any platform access limits.
4. The team's alert appetite: who gets paged, and how many false alarms per week they will tolerate (default: tune for under 2).meeting-agenda
When your agent uses it
Builds focused, time-boxed meeting agendas where every item has an owner, a timebox, and a named outcome, status updates are pushed to async, and the agenda ships 24 hours before the meeting. Use when someone asks "write an agenda for my planning meeting", "this meeting keeps running over", "how do I make this decision meeting productive", or "our recurring sync feels aimless". Do NOT use for recurring manager-report one-on-ones - use 1on1-agenda instead - or for turning meeting notes into action items - use meeting-notes-to-actions.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
1,510 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: agenda template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,422 bytes
Read SKILL.md
# Meeting Agenda Builder
An agenda is a contract for everyone's time, and the rule is absolute: no agenda, no meeting. A 60-minute meeting with 8 people spends a full person-day of work; running it on vibes wastes that person-day and, worse, trains attendees that meetings are where attention goes to die. This skill turns "let's get together" into a tight, outcome-driven agenda - or into the discovery that no meeting was needed.
## Inputs to collect
1. **Purpose** - decide, align, brainstorm, or inform. Exactly one. If the answer is "inform," stop: that is a written update, not a meeting; reclaim the hour.
2. **Attendees** - required vs. optional, with each required person's role. Anyone without a role gets cut from required. Meetings above 8 people rarely decide anything; if the list is bigger, most of them are audience - make them optional and send notes.
3. **Time budget** - total minutes. Default to 25 or 50 minutes (not 30/60) to allow transit between back-to-back calls.
4. **Decision owner** - the single person accountable for the outcome. A decision meeting without a named decider produces a discussion, not a decision.
5. **Candidate topics** - everything people want covered, raw.
## Operating procedure
meeting-notes-to-actions
When your agent uses it
Converts raw meeting notes or transcripts into a four-part record - a 3-5 sentence summary, the decisions actually made, an action table with owner, verb-phrased action, due date, and status, and the open questions - without inventing owners, dates, or tasks. Use when someone pastes notes and asks "pull the action items out of this", "summarize this meeting", "who owns what from this transcript", or "what did we actually decide". Do NOT use for preparing the agenda before a meeting happens - use meeting-agenda instead - or for turning extracted actions into engineering tickets - use jira-ticket-writer instead.
What it loads into context
Frontmatter
153 tokensAlways available to the agent
SKILL.md
1,436 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked extraction example
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,495 bytesSKILL.mdcore6,495 bytesSKILL.mdcore6,495 bytes
Read SKILL.md
# Meeting Notes to Actions
The expensive failure with meeting notes is not messiness - it is the commitment that silently evaporates because nobody wrote down who owns it. This skill extracts, rather than reformats: decisions and commitments come out of the noise, gaps in ownership are made loudly visible instead of papered over, and nothing is invented that was not in the room.
## Operating procedure
### Step 1: gather inputs
- The raw material: notes, transcript, chat log, or bullet dump. Accept it as-is.
- The meeting date if stated or supplied - relative dates like "Friday" can only be resolved to a calendar date when the meeting date is known; otherwise keep the relative form exactly as written.
- Attendee names if available, used only to disambiguate ("Alex" when two Alexes attended); never to guess owners.
If the notes are too thin to extract from (no verbs, no names), say so and return what is extractable rather than padding.
messaging-hierarchy
When your agent uses it
Use when turning a positioning statement into actual copy - building the value proposition, message pillars, proof points, and per-channel messaging that keep the website, ads, and sales deck all saying the same thing. Triggers on "write our value proposition", "message pillars", "messaging framework", "messaging house", "key messages", "proof points", "our website and sales say different things", "per-channel messaging", "on-message copy". Takes the positioning-statement as input and feeds landing-page-copy. Do NOT use when you have not yet fixed positioning - use positioning-statement first. Do NOT use when you need the actual hero/landing page words - use landing-page-copy. Do NOT use to sequence launch phases - use launch-plan-sequencer; for the hour-by-hour go-live - use launch-day-runbook; for the sales deck/talk track - use sales-enablement-kit; for the in-product activation flow - use plg-motion-designer.
What it loads into context
Frontmatter
215 tokensAlways available to the agent
SKILL.md
3,585 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The hierarchy, top to bottom
- The ordered workflow
- Quality bar
- Do NOT
- Runnable artifact - the consistency linter
- Fill-in template - the one-page messaging house
- Deliverable
Files
SKILL.mdcore14,829 bytesSKILL.mdcore14,829 bytesSKILL.mdcore14,829 bytesSKILL.mdcore14,829 bytesSKILL.mdcore14,829 bytes
Read SKILL.md
# Messaging Hierarchy
Positioning is the strategy; messaging is the words. Most teams skip the layer in
between and let every surface invent its own - so the website promises one thing,
the ads a second, and the sales deck a third, and the buyer never hears the same
story twice. This skill builds the **messaging hierarchy** (the "messaging
house"): one value proposition on top, three message pillars holding it up, proof
points under each pillar, and per-channel copy derived from - never inventing
beyond - that structure. Build it once, and every channel says the same thing in
its own register.
This skill takes the output of [[positioning-statement]] as its input and produces
the brief that [[landing-page-copy]] turns into hero and page copy. It is the
spine of the [[launch-plan-sequencer]] and the source the [[sales-enablement-kit]],meta-title-optimizer
When your agent uses it
Writes title tags and meta descriptions that survive SERP truncation, place the primary keyword early for relevance, and frame the snippet to win the click against competing results. Use when someone asks "write a title tag for this page", "why does this page rank but get no clicks", "is my title too long", "rewrite my meta description", or wants a snippet that stands out from the pattern on the results page. Do NOT use for broad on-page or technical SEO (headings, internal links, schema, crawlability, Core Web Vitals) - use seo-optimizer. Do NOT use for classifying what searchers want from a query - use search-intent-classifier. Do NOT use for finding what content the page itself is missing versus competitors - use serp-gap-analyzer.
What it loads into context
Frontmatter
180 tokensAlways available to the agent
SKILL.md
1,488 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked contrast pair
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,780 bytes
Read SKILL.md
# Meta Title Optimizer
Write the title tag and meta description for a single page so the snippet fits SERP limits, earns relevance, and wins the click. This skill owns snippet copy only; broader on-page work belongs to seo-optimizer. The costly mistake it prevents: a page that ranks but bleeds clicks to worse pages with better snippets - position without click-through is rent paid on a store nobody enters, and an overlong or generic title is the usual cause.
## Operating procedure
### Step 1: gather inputs
- The page's primary keyword - one term. A second keyword means a second page, not a longer title.
- The search intent behind it (informational, comparison, transactional). If unknown, infer and label the inference a guess, or route to search-intent-classifier.
- What the page actually delivers: its H1 and main content. The snippet must be able to keep its promise.
- The current title/description and, if the page already ranks, its click-through problem ("ranks #4, CTR under 2%").
If the keyword or the page content is unknown, ask before writing.microservices
When your agent uses it
Designs microservice architectures - service boundaries from business capabilities and team ownership, sync vs async communication choices, saga and outbox patterns for consistency, and resilience defaults - producing a boundary map with an extraction verdict per candidate service. Use when someone asks "should this be its own service", "how do I split my monolith's domains", "sync or event-driven between these services", "how do I handle transactions across services", or is reviewing a distributed architecture. Do NOT use for planning the step-by-step migration off an existing monolith - use monolith-decomposer or strangler-fig-planner for the migration process; this skill owns the target architecture. Do NOT use for designing an individual REST contract - use api-design instead.
What it loads into context
Frontmatter
164 tokensAlways available to the agent
SKILL.md
1,819 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: boundary decision - good/bad pair
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,823 bytes
Read SKILL.md
# Microservices Design
Microservices trade local complexity for distributed complexity: every boundary you draw buys independent deployment and ownership at the price of network latency, partial failure, and eventual consistency. The costly mistake this skill prevents is the distributed monolith - a system with all the operational cost of microservices and none of the independence, usually created by splitting along technical layers or by letting two services share a database.
## Operating procedure
### Step 1: Gather inputs
1. Team topology: how many teams, and who owns what today. Label guesses as guesses.
2. The bounded contexts of the domain (orders, inventory, billing…), and which change together.
3. Scaling asymmetries: which module needs 10x the resources of the rest.
4. Deploy friction evidence: how often does one team's change block another's release.
5. Operational maturity: CI/CD, observability with distributed tracing, on-call. Without these, more services means more outages.
migration-safety-checker
When your agent uses it
Rewrite a schema migration into independently deployable, zero-downtime steps that never take a long-held blocking lock on a live table. Use when you are about to run an ALTER TABLE, CREATE INDEX, add a constraint, rename a column, or change a column type against a production database with live traffic. Do NOT use for greenfield table/column design with no rows yet - use database-schema instead; do NOT use for broad correctness/integrity review of a migration - use review-db instead.
What it loads into context
Frontmatter
121 tokensAlways available to the agent
SKILL.md
986 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,485 bytes
Read SKILL.md
# Migration Safety Checker
Turn a migration that is correct in a test DB but takes a production-halting lock into a sequence of small, reversible, zero-downtime steps. The outage is rarely the DDL itself - it is the queue of fast queries that pile up behind a statement waiting for an ACCESS EXCLUSIVE lock.
## Workflow
1. **Identify the lock each statement takes.** In Postgres, ACCESS EXCLUSIVE (blocks reads AND writes) is taken by: changing a column type, adding a non-concurrent index, adding a column with a volatile default on old engines, and most constraint additions via direct ALTER. Flag every statement that takes a heavy lock on a large, write-hot table - as a rule of thumb, treat anything above a few million rows or with sustained write traffic as hot.
2. **Guard the lock acquisition.** Set `lock_timeout` (2-5 seconds is a common production setting) before any DDL so a blocked statement aborts instead of stalling traffic behind it. A long ALTER waiting for its lock blocks every query that arrives after it, including fast SELECTs - that queue is the outage.
3. **Add columns nullable, then backfill, then constrain.** Add the column NULLable with no volatile default (instant on modern Postgres). Backfill in batches of 1,000-10,000 rows (`UPDATE ... WHERE id BETWEEN ...`), committing each batch and pausing briefly between them, to avoid one giant transaction and table bloat. Add NOT NULL via a CHECK constraint with `NOT VALID` then `VALIDATE CONSTRAINT` - a weaker lock than direct `SET NOT NULL`.
4. **Build indexes and constraints without blocking.** Use `CREATE INDEX CONCURRENTLY` (cannot run in a transaction - disable the migration's transaction wrapper, e.g. `disable_ddl_transaction!` in Rails). Add foreign keys and check constraints as `NOT VALID` first (fast, light lock), then `VALIDATE CONSTRAINT` in a separate step (scans without blocking writes). Drop with `DROP INDEX CONCURRENTLY`.
5. **Use expand-contract for renames and type changes.** Never rename or retype a column in one deploy - running code still references the old shape. Expand: add the new column, dual-write from the app, backfill. Migrate reads to the new column. Contract: stop writing the old column, then drop it in a later deploy. Each step is independently deployable and reversible.
6. **Verify the actual lock before shipping.** Confirm the rewritten statement's lock with a lock-impact migration linter (e.g. squawk) or by inspecting `pg_locks` while the statement runs on a staging clone - never assume from the SQL alone.
## Deliverable
ml-feature-engineering
When your agent uses it
Designs ML features with leakage-safe pipelines, correct categorical encoding by cardinality, numeric transforms, and validation that a feature earns its place. Use when someone asks "how should I encode this high-cardinality column", "why does my model score great offline and fail in production", "what features should I build from this table", or "should I scale these inputs". Do NOT use for building the serving infrastructure that stores and versions features across models - use feature-store-design instead; for detecting when live feature distributions shift after deployment - use data-drift-monitor instead; for initial dataset exploration and profiling - use eda-playbook instead.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
1,703 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: leakage contrast pair
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,962 bytesSKILL.mdcore7,962 bytesSKILL.mdcore7,962 bytesSKILL.mdcore7,962 bytes
Read SKILL.md
# Feature Engineering
Most "amazing offline, useless in production" models are not modeling failures - they are leakage failures introduced during feature engineering. This skill builds features that improve cross-validated performance without smuggling the target into the inputs, and rejects features that only look good because they cheated.
## Operating procedure
Leakage prevention comes first because every later step (encoding, aggregation, imputation) is a leakage opportunity; validation comes last because a feature that does not move the cross-validated metric does not ship.
### Step 1: Gather inputs
Collect before writing transforms; label unknowns as guesses.
1. Prediction task and the exact prediction time: what is known at the moment the model must predict? Everything else is off-limits.
2. Target definition and how it was produced - features derived from the same process as the label are proxy-leakage suspects.mobile-offline-sync
When your agent uses it
Builds local-first mobile storage with an optimistic local store, durable mutation outbox, per-entity conflict-resolution rules, incremental pull, and idempotent background retry. Use when someone says "the app has to work offline", "how do I sync Room or SQLite or Realm with the server", "what happens when two devices edit the same record", or offline edits must reconcile with the server later. Do NOT use for paginating, caching, and refreshing read-heavy server lists - use pagination-and-sync-engineer instead; and do NOT build offline write support for strong-consistency operations like payments or inventory counts - disable those offline.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
1,657 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: conflict-resolution decision for a field-service app
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore8,004 bytesSKILL.mdcore8,004 bytes
Read SKILL.md
# Mobile Offline Sync
Make the local store the source of truth the UI reads and writes, and treat the network as eventual replication. The hard part is never the happy path - it is reconciling divergent edits made while offline. The costly failure is an ad-hoc "retry the request later" queue that double-applies mutations, resurrects deleted rows, and silently destroys one user's edits with another's.
## Operating procedure
### Step 1: Gather inputs
Collect per feature before designing:
1. The entities that must work offline, and for each: single-owner or multi-writer? Whole-object edits or field-level edits?
2. The maximum realistic offline window (a field-service app may need 7-30 days; a consumer app usually hours). This sets tombstone retention.
3. The local store (Room, SwiftData, SQLite, Realm) and whether the server API can accept client-generated UUIDs and return per-record versions.
4. Which operations are strong-consistency (payments, inventory decrements, seat booking) - these get disabled offline, not synced.
5. Expected data volume per user, which decides pull batch sizes.mobile-perf-profiler
When your agent uses it
Diagnoses and fixes mobile jank, dropped frames, slow startup, and growing memory by capturing a trace or heap snapshot on a real device, isolating the single worst cost, fixing it, and re-measuring against the 16ms frame and cold-start budgets. Use when someone says "the app stutters when I scroll", "startup takes forever", "memory keeps climbing after navigating around", or the UI feels laggy on a real device. Do NOT use when the bottleneck is a website or web app (LCP, bundle size, slow endpoint or query) - use web-performance instead; for Compose recomposition fixes pair with jetpack-compose-builder; for Flutter widget architecture use flutter-widget-architect.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,835 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Budgets and red lines
- Worked artifact: trace-reading decision example
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore8,162 bytesSKILL.mdcore8,162 bytesSKILL.mdcore8,162 bytes
Read SKILL.md
# Mobile Perf Profiler
Diagnose mobile performance from measured evidence, fix the largest cost, and re-measure - never guess. The costly failure is folklore optimization: weeks spent caching and micro-tuning paths the profiler would show are cheap, while the one 40ms bind on the scroll path ships untouched.
## Operating procedure
### Step 1: Gather inputs
1. The exact symptom, named precisely: jank while scrolling a specific list, jank on first frame of a screen, slow cold start, or memory that climbs. "The app feels slow" is not a symptom yet - reproduce until it is one.
2. A representative device: a real mid-tier phone from the install base, not the developer's flagship and never a simulator/emulator.
3. A release or profile build. Debug builds misreport timing by integer factors.
4. The display refresh rate of target devices - 60Hz means a 16.6ms frame budget, 120Hz means 8.3ms.
5. Field data access: Play Console Android vitals, Xcode Organizer / MetricKit. If available, this decides which symptom is worth the trace before any local work.
### Step 2: Work the evidence ladder, cheapest firstmobility-routine
When your agent uses it
Builds a 10-15 minute daily floor-based mobility routine for desk workers, targeting the four zones sitting predictably stiffens - hips, thoracic spine, shoulders/neck, and ankles - with exact holds, reps, and a sequenced routine card. Use when someone says "I'm stiff from sitting all day", "my hips are tight", "fix my desk posture", or "build me a daily stretching routine". General movement guidance, not medical advice. Do NOT use for injury-prevention work attached to a training program - use injury-prehab instead; for a lifting program use strength-training-plan; for HIIT or conditioning use fitness-program.
What it loads into context
Frontmatter
149 tokensAlways available to the agent
SKILL.md
1,680 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,415 bytes
Read SKILL.md
# Mobility Routine
Desk workers do not get randomly stiff - 6-10 seated hours produce one predictable pattern: hip flexors shorten, glutes underactivate, the thoracic spine rounds and stiffens, the head juts forward, shoulders roll inward, and ankle dorsiflexion quietly degrades. A routine that ignores this pattern (generic toe-touches, aggressive neck stretching) wastes the ten minutes it takes. This skill builds a 10-15 minute daily sequence aimed exactly at those zones, balancing lengthening with loaded end-range reinforcement so the new range actually sticks.
## Operating procedure
### Step 1: Gather inputs
Collect: hours seated per day; where stiffness is felt most (hips, mid-back, neck/shoulders, ankles - bias 40% of routine time toward the worst zone); minutes genuinely available daily (default 12-15; floor at 8); preferred slot (morning or post-work - both work; pick whichever survives the user's real schedule); equipment (a foam roller helps the thoracic work but a rolled towel substitutes); and any current pain. Sharp, radiating, or numbness-type pain excludes that zone and routes to Escalation. If the user cannot localize their stiffness, treat the hip complex as the priority - it is the highest-return zone for sitters.
### Step 2: Program the hip complex
The 90/90 is the highest-return single mobility exercise for desk workers: both legs at 90 degrees, one in front and one to the side; hold the forward-shin position 60-90 seconds per side, gently rotating the torso toward the front leg. Follow with the couch stretch - rear foot elevated, lunge position, deliberate posterior pelvic tilt - 60 seconds per side. Between them these cover the hip flexors and both internal and external hip rotation. The posterior tilt is the technique detail that decides whether the couch stretch reaches the hip flexor or just cranks the lower back; cue it explicitly.
mock-stub-designer
When your agent uses it
Designs the minimal set of test doubles for a unit or integration test and decides, per dependency, whether to stub, fake, spy, mock, or exercise the real thing. Use when someone asks "should I mock this", "how do I test code that calls Stripe or S3 or the clock", "why do our tests pass while production is broken", or when a test touches an HTTP API, payment or email/SMS SDK, database, filesystem, system clock, or randomness. Do NOT use for fabricating valid domain fixture data (a User with defaults, an Order with line items) - use test-data-builder instead; for driving a red-green implementation loop, use tdd-expert.
What it loads into context
Frontmatter
161 tokensAlways available to the agent
SKILL.md
1,711 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: good vs bad double design
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore8,028 bytesSKILL.mdcore8,028 bytesSKILL.mdcore8,028 bytes
Read SKILL.md
# Mock Stub Designer
A test double exists to make a test fast, deterministic, and focused - never to mock everything in sight. Over-mocking yields tautological tests that pass against a broken system because they only verify the mocks. This skill decides, per dependency, the weakest double that still isolates the boundary, so the test exercises real behavior everywhere else.
## Operating procedure
### Step 1: Gather inputs
Collect before designing anything:
1. The unit under test and the behavior being verified (one sentence).
2. Every dependency the code touches, direct and transitive-at-the-boundary: HTTP/third-party APIs, payment/email/SMS SDKs, database, filesystem, clock, randomness, queues.
3. Language and test stack (this picks the wire-interception tool: nock or msw for JS, WireMock for JVM, VCR-style cassettes for Ruby/Python, responses/respx for Python).
4. Whether a sandbox environment exists for the critical integration backstop.
model-card-writer
When your agent uses it
Produces a complete model card - intended use, training data provenance, evaluation results with slices, limitations, and usage guidance - for any model shared beyond its team or affecting people. Use when someone asks "write a model card", "document this model before release", "what should our model documentation include", or is preparing a model for deployment, handoff, or external publication. Do NOT use for producing the underlying evaluation numbers - use model-evaluation-report instead; for ongoing production monitoring plans use data-drift-monitor.
What it loads into context
Frontmatter
123 tokensAlways available to the agent
SKILL.md
1,452 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: filled intended-use section
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,272 bytes
Read SKILL.md
# Model Card Writer
A model card is the contract between the team that built a model and the people who will use or be affected by it. It documents not just what the model does but what it must not be used for - and the cost of skipping it is a model quietly repurposed for a use case it was never validated on, discovered only after harm. This skill produces a card a consumer can act on, not a compliance checkbox.
## Operating procedure
Write sections in the order below; intended use (Step 2) constrains everything else, and limitations (Step 5) cannot be written honestly until evaluation results (Step 4) are in front of you.
### Step 1: Gather inputs
Collect before writing; label anything unverified as a guess and chase it before publishing.
1. Model name, version, and owning team or contact.
2. Architecture family in plain language (gradient boosted trees, transformer, logistic regression).model-evaluation-report
When your agent uses it
Produces a rigorous, honest evaluation report for an ML model - real baselines, business-matched metrics with confidence intervals, slice-level error analysis, calibration checks, and a go/no-go recommendation. Use when someone asks "is this model good enough to ship", "write up the eval for this model", "did we actually beat the old model", or before any model promotion, stakeholder demo, or retraining decision. Do NOT use for evaluating LLM or prompt outputs - use llm-evaluation instead; for analyzing randomized product experiments use ab-test-analyzer; for documenting a shipped model for consumers use model-card-writer.
What it loads into context
Frontmatter
146 tokensAlways available to the agent
SKILL.md
1,603 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: executive summary and metric table
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,549 bytes
Read SKILL.md
# Model Evaluation Report
A model that beats a naive baseline on aggregate accuracy can still be worse than useless in production. Honest evaluation means choosing the right metric before looking at results, comparing against baselines that actually matter, and finding the slices where the model fails - because those slices are where the incident reports will come from.
## Operating procedure
Order matters: the primary metric and baselines are chosen before results are examined, or the report is p-hacked by construction.
### Step 1: Gather inputs
Collect from the requester before computing anything; label unknowns as guesses.
1. Task type: classification, regression, or ranking.
2. The business decision the model drives, and the asymmetry of errors (is a false positive or a false negative more expensive, and roughly by how much).mongodb-expert
When your agent uses it
Designs MongoDB schemas, indexes, and aggregation pipelines that perform - embed-vs-reference decision rules, the 16MB document limit, ESR compound-index ordering, explain-plan verification, and operational settings for replica sets and sharding. Use when someone asks "should I embed or reference this", "why is my Mongo query slow", "design a MongoDB schema for X", "how do I structure this aggregation pipeline", or "what should my shard key be". Do NOT use for relational or SQL schema design - use database-schema instead; for tuning SQL queries use sql-query-optimizer; for Postgres-on-Supabase work use supabase-expert.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,978 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: schema pair, bad vs good
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore8,805 bytes
Read SKILL.md
# MongoDB Expert
Model data, index it, and query it the MongoDB way: around access patterns, not normalization. The costly mistake this skill prevents is a relationally-normalized schema ported into Mongo - a design that forces application-side joins on every read and, in the opposite failure, unbounded embedded arrays that grow until documents hit the 16MB limit and writes start failing in production.
## Operating procedure
Schema first, indexes second, pipelines third - indexes serve queries, and queries are determined by the schema, so working backwards wastes every downstream decision.
### Step 1: Gather inputs
1. The top read and write patterns, ranked by frequency: which entities are fetched together, filtered by what, sorted by what.
2. Cardinality and growth of every one-to-many relationship: bounded ("an order has 5-50 line items") or unbounded ("a user accrues events forever"). Label estimates as guesses.
3. Read/write ratio per collection and expected document counts.
4. Deployment shape: replica set or sharded cluster, and durability requirements.monolith-decomposer
When your agent uses it
Finds and validates one bounded-context seam to extract from a monolith - gated on coupling-graph, co-change, data-ownership, and transaction-boundary evidence - and sequences an incremental strangler-fig extraction plan. Use when someone asks "where should we cut this monolith", "is this module ready to extract as a service", "plan the extraction of billing from the app", or is evaluating whether a candidate seam is clean. Do NOT use for designing the target service architecture, service communication, or greenfield service boundaries - use microservices instead; do NOT use to document the implicit business rules inside the code being moved - use business-rule-extractor instead. This skill decides where to cut and in what order.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,789 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: seam evidence summary
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,538 bytes
Read SKILL.md
# Monolith Decomposer
Extract one bounded-context seam at a time, only after data ownership and coupling evidence prove the cut is clean. Cut along the wrong seam and you get a distributed monolith: the latency, failure modes, and operational cost of microservices with the coupling of a monolith - the single most expensive architecture mistake a team can make.
## Operating procedure
Evidence gates come before design, and design before sequencing, because a seam that fails the data-ownership veto in Step 3 makes every downstream step wasted work.
### Step 1: Gather inputs and confirm the motive
Collect: the candidate context (if any), the driver (independent scaling, deploy cadence, team ownership - the only three valid ones), team size, and current deploy pain. If the driver is slow deploys, flaky tests, or unclear modules, stop: modularize in-process first. A well-modularized monolith beats a badly-cut set of services on nearly every axis. Label an unverified driver as a guess and verify it before proceeding.
### Step 2: Pick a candidate bounded context
Name one cluster of behavior with high internal cohesion and a thin, stable interface to the rest - a domain concept (DDD bounded context / aggregate), not an org-chart team or a technical layer. Extract exactly one per effort.monorepo-expert
When your agent uses it
Structures and scales JavaScript/TypeScript monorepos with Turborepo, Nx, and pnpm workspaces - layout rules, task pipelines, remote caching, and affected-only CI so build times stay flat as the repo grows. Use when someone asks "should we use a monorepo", "how do I set up Turborepo or Nx", "our monorepo CI takes 40 minutes", "how do I share code between apps", or "why is the cache never hitting". Do NOT use for authoring the CI workflow files themselves - use github-actions instead - and do NOT use for deciding microservice boundaries, which is about runtime architecture, not repo layout - use microservices instead.
What it loads into context
Frontmatter
166 tokensAlways available to the agent
SKILL.md
1,880 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: good/bad pipeline pair
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,895 bytesSKILL.mdcore7,895 bytes
Read SKILL.md
# Monorepo Expert
A monorepo's promise is shared code with atomic cross-package changes; its failure mode is CI time growing linearly with repo size until every PR waits 40 minutes to test a one-line change. The costly mistake this skill prevents is running every task on every change - the entire discipline of monorepo tooling reduces to one rule: build and test only what the diff affected, and cache everything else.
## Operating procedure
### Step 1: Gather inputs
1. Package manager and tool already in use (pnpm/npm/yarn; Turborepo/Nx/none). Default recommendation for a fresh setup: pnpm + Turborepo for simplicity, Nx when you want generators, enforced boundaries, and plugin depth.
2. Number of apps and packages, and expected growth.
3. Current CI wall-clock time and what a PR actually runs today.
4. Whether builds are reproducible - any task reading undeclared env vars or the network.
5. Team count - boundaries matter more with more teams.
month-end-close
When your agent uses it
Guides finance teams through a controlled month-end close - sub-ledger cutoffs, journal entries in dependency order, full balance-sheet reconciliation, flux analysis, and sign-off - targeting a locked close by business day 5-10. Use when someone asks "help me close the books", "build a close checklist", "our close takes three weeks, how do we shorten it", "what order do the journal entries go in", or is preparing for an audit. Do NOT use for constructing or interpreting the financial statements themselves - use financial-statement-builder instead - or for writing the budget-variance narrative that follows the close - use budget-vs-actual instead.
What it loads into context
Frontmatter
151 tokensAlways available to the agent
SKILL.md
1,858 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: close checklist ordered by dependency
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,942 bytesSKILL.mdcore7,942 bytesSKILL.mdcore7,942 bytes
Read SKILL.md
# Month-End Close
A fast, clean close is the result of consistent preparation, not heroics at deadline. The failure this skill prevents is the restatement: a late invoice posted after reconciliation, an unreconciled variance carried forward "to fix next month," a flux swing nobody explained that turns out to be a misposted entry. The sequence below has hard checkpoints so nothing slips and the books reflect economic reality.
Target: sign-off and lock by business day 5 for a well-run close; day 10 is the outer acceptable bound. A close consistently past day 10 means the pre-close phase is being skipped - fix the calendar, not the deadline.
## Operating procedure
The order is dependency order, and it is not negotiable: journals before sub-ledger lock, lock before reconciliation, reconciliation before flux, flux before sign-off. Reconciling before the lock guarantees rework; running flux before reconciliation means explaining errors instead of catching them.
### Step 1: Gather inputs (once, when adopting this process)
1. Chart of accounts and entity structure (any intercompany?).
2. Sub-ledger systems for AP, AR, payroll, inventory, and who owns each cutoff.moodboard-builder
When your agent uses it
Assembles a curated moodboard of 8 to 16 images plus a 150-300 word art direction statement that argues for one visual direction. Use when someone asks "build a moodboard", "set the visual direction for this brand", "get the team aligned on look and feel before design starts", or is presenting art direction to a client. Do NOT use for sourcing and licensing individual production images - use visual-asset-curation or stock-photo-finder instead; for codifying an already-approved identity into usage rules, use brand-guidelines.
What it loads into context
Frontmatter
130 tokensAlways available to the agent
SKILL.md
1,505 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template: art direction statement
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,036 bytesSKILL.mdcore7,036 bytes
Read SKILL.md
# Moodboard Builder
A moodboard is not a Pinterest dump - it is a curated argument for one specific visual direction, made before pixels are pushed. The costly failure mode is a board of 40 nice-looking images that lets every stakeholder see whatever they want in it, so "alignment" evaporates the moment real design work appears. This procedure produces a tight board and a written statement that survive the client meeting.
## Operating procedure
Order matters: the emotional target is the filter every image must pass through, so writing it after sourcing guarantees a board justified backwards.
### Step 1: Gather inputs
Collect before sourcing anything. Where the user does not know, use the default and label the guess as a guess.
- What the brand does and who the primary customer is (no default - ask).
- Project stage: new brand, rebrand, or refresh (changes how far the board may stray from existing equity; default: new brand).motion-color-and-light
When your agent uses it
Choose palette, contrast, and lighting for a moving video - not a static UI - so color survives H.264/VP9 compression, banding, and YouTube/Instagram/TikTok recompression across OLED, LCD, and projectors. Use when picking video colors, building a gradient mesh, adding glow, bloom, or depth, setting a dark-background default, or asking "why does my gradient band", "why is my neon muddy after upload", "is this contrast video-safe", or "what palette for my video". Do NOT use for scene timing (use video-storyboard), easing and animation feel (use motion-design-principles), styling on-screen text (use kinetic-typography), structuring a product walkthrough (use product-demo-director), audio (use sound-and-music-sync), or reframing for a platform (use social-video-formatter).
What it loads into context
Frontmatter
200 tokensAlways available to the agent
SKILL.md
4,249 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Template: color-and-light tokens
- references/why-video-color-is-different
- references/dark-vs-light-stage
Files
SKILL.mdcore16,286 bytesSKILL.mdcore16,286 bytesSKILL.mdcore16,286 bytesSKILL.mdcore16,286 bytes
Read SKILL.md
# Motion Color and Light
This is the taste layer for color in motion. A palette that looks immaculate as a
static Figma frame can fall apart the moment it moves and gets encoded: smooth
gradients break into visible stair-step bands, saturated neon turns to muddy
mush, and a contrast ratio that passed for UI text disappears under motion blur.
Video is not UI. The pixels you author are not the pixels the viewer sees - they
pass through H.264/VP9 compression, chroma subsampling, and a second
re-compression pass when YouTube, Instagram, or TikTok ingests the file, then land
on a screen you do not control (a bright OLED phone, a washed-out conference
projector, an old LCD).
You decide palette, contrast, and lighting so the image survives all of that.
This pairs with the `remotion-video-production` mechanics pack: you make the colormotion-design-principles
When your agent uses it
The craft layer for motion: pick the right easing (ease-in, ease-out, ease-in-out, spring), get timing and spacing right, build the anticipation-action-follow-through arc, and apply Disney's 12 principles to UI and product video. Use when animating any UI transition, modal, button, page load, or product-video scene, or when someone says "make this feel less janky", "smoother", "more polished", "the animation feels off", "why does this look cheap", or asks "ease-in or ease-out" / "spring or interpolate". Do NOT use for shot planning (use video-storyboard), directing a product demo (use product-demo-director), animating text (use kinetic-typography), timing to music (use sound-and-music-sync), platform reframing (use social-video-formatter), or color and light (use motion-color-and-light).
What it loads into context
Frontmatter
207 tokensAlways available to the agent
SKILL.md
5,117 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use a sibling instead
- Operating procedure
- Quality bar
- Do NOT
- Easing & spring picker
- Template: motion-spec
- references/twelve-principles
- references/spring-vs-interpolate
Files
SKILL.mdcore19,917 bytesSKILL.mdcore19,917 bytesSKILL.mdcore19,917 bytesSKILL.mdcore19,917 bytes
Read SKILL.md
# Motion Design Principles
This is the taste layer. It decides *how* something should move so it reads as
polished and intentional instead of janky, floaty, or cheap. It does not write
the code that plays the motion â<U+0080><U+0094> once you have decided the curve, the timing, and
the arc, that decision is the *input* to a directing/render skill (for product
video, hand it to **product-demo-director**) or to your CSS/JS transition (for
UI). Keep the two jobs separate: this skill owns the judgment; the directing
skill owns the orchestration and mechanics.
Fire on anything about how motion *feels*: "make this less janky", "smoother",
"more polished", "more alive", "the easing is off", "this looks cheap", "should I
ease-in or ease-out", "spring or interpolate".
motion-spec
When your agent uses it
Specifies animation and motion precisely - duration, easing curve, trigger, and intent - for accurate engineer implementation. Use when handing off transitions, micro-interactions, or loading states.
What it loads into context
Frontmatter
51 tokensAlways available to the agent
SKILL.md
987 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The Five Required Fields
- Standard Duration Reference
- Easing Conventions
- Orchestration and Sequencing
- Reduced Motion
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore4,675 bytes
Read SKILL.md
# Motion Spec
Motion that is vaguely specified gets implemented incorrectly or skipped. Every animation needs five pieces of information: what moves, when it triggers, how long it takes, what easing curve it follows, and why it exists. Provide all five.
## The Five Required Fields
For every animated element, document:
1. Element: the component or property being animated (e.g. 'modal overlay opacity', 'button background color').
2. Trigger: the event that starts the animation (e.g. 'user clicks confirm', 'data fetch resolves', 'component mounts').
3. Duration: in milliseconds. Do not use vague terms like 'fast' or 'subtle'.
4. Easing: the named curve or cubic-bezier values (e.g. 'ease-out', 'cubic-bezier(0.4, 0, 0.2, 1)'). Note if the design tool curve must be converted.
5. Intent: one sentence on why the motion exists - it guides engineers when they must adapt the spec to constraints.
## Standard Duration Referencemutation-test-runner
When your agent uses it
Runs mutation testing on already-covered code and turns each surviving mutant into a specific missing assertion, exposing tests that execute code but verify nothing. Use when line coverage is high yet a bug slipped through, when reviewing a critical module (pricing, permissions, auth, state machines) before release, or when you are asked to run Stryker, PIT, mutmut, or cosmic-ray. Do NOT use when the goal is to find which code paths are untested at all - use coverage-gap-finder instead.
What it loads into context
Frontmatter
121 tokensAlways available to the agent
SKILL.md
747 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
Files
SKILL.mdcore3,404 bytes
Read SKILL.md
# Mutation Test Runner
Prove whether covered code is actually verified: deliberately break a line, and check that a test notices. A surviving mutant is proof that the line is executed but unasserted.
## Workflow
1. Pick the target. Scope to high-risk logic where a silent bug is expensive - pricing, permissions, auth, state machines, financial math. Skip generated code, trivial DTOs, and glue; they have nothing worth mutating.
2. Pick the runner: Stryker (JS/TS/C#), PIT (JVM), mutmut or cosmic-ray (Python). Configure it to run against the target's existing test suite.
3. Run scoped, not global. In CI, mutate only changed files (Stryker incremental / diff mode); reserve a full sweep for a nightly job. Enable coverage analysis so only tests touching a mutated line rerun, and parallelize across cores.
4. Read each surviving mutant as a named missing assertion, not a vague gap:
- `<` mutated to `<=` survives -> the boundary case is unasserted; add an assertion at the exact edge value.
- A removed function call survives -> no test checks that side effect; assert the side effect.
- A negated condition survives -> the two branches are indistinguishable to the suite; assert the differing outcome of each branch.
- A return value replaced with a constant survives -> the return is unchecked; assert it.
5. Triage equivalent mutants. Some mutants cannot change observable behavior (a change in dead code, `i++` vs `++i` in isolation). Distinguishing them is undecidable in general - mark them ignored in the tool's config rather than chasing 100%.mutual-action-plan
When your agent uses it
Builds a mutual action plan (MAP) with shared milestones, named owners, and buffer-padded dates that drives a qualified deal from verbal intent to signature and go-live. Use when someone asks "build a mutual action plan for this deal", "how do I keep this deal from stalling", "the buyer said yes but nothing is moving", or once a prospect is qualified and has expressed intent to move forward. Do NOT use for preparing the qualification call itself - use discovery-call-prep instead. Do NOT use for scripting the negotiation and close conversation - use closer-sales-script instead.
What it loads into context
Frontmatter
136 tokensAlways available to the agent
SKILL.md
1,803 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Template: MAP skeleton
- Deliverable
- Do NOT
- Quality bar
- Escape hatch
Files
SKILL.mdcore8,321 bytes
Read SKILL.md
# Mutual Action Plan
A mutual action plan is a shared document that aligns buyer and seller on every step between "we want to move forward" and "the contract is signed and live." The word mutual matters: if the buyer has not contributed to and agreed on the plan, it is a vendor's wishlist, and the deal it was supposed to protect will stall in legal, procurement, or a budget cycle nobody mentioned. The costly mistake this skill prevents is discovering a six-week security review three weeks before the quarter ends.
## Operating procedure
Steps run in this order because dates cannot be set before milestones are known, and milestones cannot be known before the buyer has contributed theirs.
### Step 1: Gather inputs
Collect before drafting anything. Label guesses as guesses.
1. Deal stage and evidence of intent - a MAP before clear evaluation intent is premature. Default assumption: qualified with verbal intent.
2. Buyer's desired go-live date, or the business event driving urgency. If neither exists, that absence is a finding (see Step 3).n-plus-one-hunter
When your agent uses it
Detect ORM loop-of-queries (N+1) patterns from query logs and eliminate them with batched eager loading, keeping the N+1s that are cheap. Use when a list endpoint is slow, a request issues many near-identical SELECTs differing only by an id, query logs show repeated SELECTs in a loop, or a tool like the bullet gem flags an N+1 in ActiveRecord, Prisma, SQLAlchemy, or Hibernate.
What it loads into context
Frontmatter
113 tokensAlways available to the agent
SKILL.md
1,211 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,002 bytes
Read SKILL.md
# N+1 Query Hunter
Find ORM N+1 patterns from captured query logs and replace them with batched eager loading, leaving cheap bounded ones alone. An N+1 is one query to fetch a list, then one more query per row to fetch its association - the most common ORM performance bug, hidden because each query is individually fast.
Do NOT use when reading or interpreting a single statement's EXPLAIN/EXPLAIN ANALYZE plan - use explain-plan-reader instead. Do NOT use when the fix is restructuring one query's SQL (rewriting joins, subqueries, or window functions) - use query-rewriter instead. This skill owns the loop-of-queries case where the fix is preloading, not rewriting one statement.
## Workflow
1. Capture the evidence before guessing. Turn on query logging and count the SELECTs for the slow request. ActiveRecord: tail the log or add the `bullet` gem. SQLAlchemy: `create_engine(..., echo='debug')`. Prisma: enable the `query` log event. Hibernate: `hibernate.show_sql=true` and `generate_statistics=true`. The tell is a burst of near-identical SELECTs differing only in the `WHERE id = ?` value, scaling with collection size - 20+ of them in one request is a near-certain N+1. Confirm server-side with `pg_stat_statements`: a normalized query whose `calls` count is orders of magnitude above the endpoint's request rate.
2. Confirm it scales with input, and that it's worth fixing. Note how N grows: with collection size, page size, or user-controlled input. A useful bar: fix it when the request issues 10+ queries that grow linearly with rows, or when the loop's aggregate time is 20%+ of request latency - each query may cost only 0.5-1ms, but 200 of them adds 100-200ms plus per-query round-trip overhead. A 2-query "N+1" or one bounded to a handful of rows is not the target - move on.
3. Replace lazy per-row loads with one batched preload. ActiveRecord: `Post.includes(:author, comments: :user)` - `preload` forces separate batched queries, `eager_load` forces a LEFT JOIN, `includes` lets Rails choose. SQLAlchemy: `selectinload(Post.comments)` (second IN query, best default for collections) or `joinedload` for one-to-one. Prisma: pass `include`/`select` with nested relations in one call. Hibernate: `JOIN FETCH` in JPQL or an `@EntityGraph`; never rely on global `FetchType.EAGER`. If a batched IN list can exceed roughly 1,000 ids, chunk it - huge IN lists degrade planning and can hit protocol parameter limits.
4. Avoid the fan-out trap. `joinedload`/`eager_load` on a one-to-many multiplies rows (cartesian fan-out) and can be slower than the N+1 it replaced - prefer `selectinload`/`preload` for collections. Eager loading inside a method that is itself called in a loop just moves the N+1 up one level; hoist the preload to the outermost collection.
5. Use counts, not rows, for sizes. If the code loads children only to call `.size`/`.count`, use `counter_cache` or a grouped `COUNT` instead of materializing the association.
6. Re-measure. Re-run the request with logging on and confirm the SELECT count dropped to a constant, and that p95 latency improved on the hot path.
narration-script
When your agent uses it
Write the spoken voiceover for a tutorial or demo - the actual words, phrased for the ear and synced one-instruction-per-action to what is on screen, with a pacing budget and a confident, jargon-checked tone. Use when someone says "write the voiceover", "script the narration", "what should I say over this", "VO script for my tutorial", "narrate this demo", "the script feels robotic", or "the narration is too wordy". Do NOT use to mix, time, duck, source, or beat-match the audio - VO leveling, SFX, music - that is sound-and-music-sync; do NOT use to decide which scenes exist or their order - that is video-storyboard; do NOT use to direct the on-screen actions themselves (cursor, zoom, callouts) - that is product-demo-director; do NOT use to turn spoken lines into animated on-screen text - that is kinetic-typography.
What it loads into context
Frontmatter
213 tokensAlways available to the agent
SKILL.md
971 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Write for the ear, not the page
- Sync narration to action
- Pace it
- Structure each step
- Tone
- Output format
- Don't
Files
SKILL.mdcore4,230 bytes
Read SKILL.md
# Narration Script
A tutorial lives or dies on its narration. The viewer is trying to *do* the thing while listening, so every word competes with their attention on the screen. Write less, say it plainer, and pin each line to a visible action.
## Write for the ear, not the page
Spoken language is not written language. Read every line aloud before keeping it.
- Short sentences. One idea each. If you run out of breath, it is too long.
- Active voice, present tense, second person: "Click **Save**" - not "The Save button would then be clicked by the user."
- Cut every "basically", "simply", and "just." "Just click Settings" quietly tells a stuck viewer they are stupid. Drop it.
- Contractions are good ("you'll", "it's"); they sound human.
## Sync narration to actionnetworking-system
When your agent uses it
Builds a repeatable professional networking system - a tiered contact tracker, give-first outreach and reconnect scripts, a follow-up cadence by tier, and a weekly relationship-hour routine - so staying in touch stops depending on willpower. Use when someone says "help me network for my job search", "how do I reach out to someone I admire", "I'm bad at keeping in touch with contacts", or wants a lightweight personal CRM habit. Do NOT use for sales prospecting sequences - use outreach-sequence-designer instead; for personalizing recruiting outreach, use candidate-outreach-personalizer.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
620 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- 1. Define goals
- 2. Map your network in tiers
- 3. Lead with generosity
- 4. Outreach templates
- 5. Run good conversations
- 6. Follow-up cadence
- 7. Track and remember
- 8. Maintain, don't only mine
- Output for the user
Files
SKILL.mdcore2,788 bytes
Read SKILL.md
# Networking System
Use this skill to turn networking from awkward, sporadic effort into a calm, repeatable system built on genuine relationships.
## 1. Define goals
Clarify why you are networking: job search, learning, partnerships, mentorship, or visibility. Goals shape who you reach out to and what you ask for.
## 2. Map your network in tiers
- Tier 1: close contacts you can reach anytime.
- Tier 2: warm acquaintances worth nurturing.
- Tier 3: aspirational contacts you want to meet.
Keep a simple CRM (spreadsheet or contacts app) with name, context, last touch, and next action.
## 3. Lead with generositynext-on-vercel-perf
When your agent uses it
Diagnose and fix Core Web Vitals and page performance of a Next.js app on Vercel - measurement-first. Use when an LCP/INP/CLS score is bad, a page "feels slow", Lighthouse/PageSpeed/Speed Insights is red, the JS bundle is too big, images or fonts are janky, or you want Vercel-specific caching/edge wins. Triggers: "improve Core Web Vitals", "fix LCP", "reduce INP", "stop layout shift", "my Next.js page is slow on Vercel", "shrink my bundle", "next/image optimization", "font is causing CLS", "cache this page". Do NOT use to set up the deploy pipeline or rollbacks - use vercel-deploy-pipeline; for ISR/cache-component semantics and edge runtime choice - use vercel-edge-and-isr; for env/secrets - use vercel-env-management; for model/provider routing - use vercel-ai-gateway; for WAF/bot defense - use vercel-firewall-and-botid.
What it loads into context
Frontmatter
245 tokensAlways available to the agent
SKILL.md
4,663 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Step 1 - Get a real measurement before touching code
- Step 2 - Route the failing metric to its root cause
- Step 3 - Images: the #1 LCP and CLS lever
- Step 4 - Bundle size: the #1 INP lever
- Step 5 - Fonts: a quiet LCP + CLS tax
- Step 6 - Vercel caching: cut TTFB so LCP starts sooner
- Quality bar
- Do NOT
- Calculator
- Template: perf-audit-record
Files
SKILL.mdcore16,958 bytesSKILL.mdcore16,958 bytesSKILL.mdcore16,958 bytesSKILL.mdcore16,958 bytesSKILL.mdcore16,958 bytesSKILL.mdcore16,958 bytes
Read SKILL.md
# Next on Vercel Performance
You make a Next.js app on Vercel measurably faster - by **measuring first**, fixing
the one metric that is actually red, and verifying the number moved. Performance
work without a number in front of you is guesswork; this skill refuses to guess.
Trigger eagerly on any signal that a page is slow or a vital is failing: "fix my
LCP", "reduce INP", "stop the layout shift", "Lighthouse is red", "Speed Insights
is bad", "my bundle is huge", "the page feels janky on mobile", "make this page
faster on Vercel". Each Core Web Vital has a *different* root cause and a
*different* fix - never apply an INP fix to an LCP problem. This skill is one stop
in the ship-a-Next.js-app-on-Vercel workflow: it assumes the app already deploys
(vercel-deploy-pipeline) and that ISR / runtime choices live in
vercel-edge-and-isr; here we tune what those produce.nextjs-app-router
When your agent uses it
Builds and reviews Next.js App Router code - server/client component boundaries, data fetching, caching and revalidation choices, streaming with Suspense, and Server Action mutations - and delivers routes where every dynamic-vs-cached decision is explicit. Use when someone asks "should this be a client component", "why is my data stale", "where do I put use client", "how do I stream this page", or "how do server actions work". Do NOT use for Vercel-specific runtime and cost tuning of an already-built app - use next-on-vercel-perf instead; for edge middleware and ISR deployment topology - use vercel-edge-and-isr instead; for framework-agnostic Core Web Vitals work - use web-performance instead; for designing an application-wide cache layer beyond the framework - use caching-strategy instead.
What it loads into context
Frontmatter
187 tokensAlways available to the agent
SKILL.md
1,678 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,448 bytesSKILL.mdcore7,448 bytesSKILL.mdcore7,448 bytes
Read SKILL.md
# Next.js App Router
Default to Server Components; reach for Client Components only for interactivity, state, or browser APIs. The two costly mistakes this skill prevents are the accidental client tree - one `'use client'` placed high that drags the whole page into the browser bundle - and accidental caching decisions, where a team assumes data is cached (or fresh) without checking which caching model the app runs on. Since Next.js 15, `fetch` is not cached by default; caching is something you opt into, explicitly.
## Operating procedure
### Step 1: gather inputs
Collect before writing a route (label guesses as guesses):
0. The Next.js version (`next` in package.json) and whether `cacheComponents: true` is set in next.config. The caching model below depends on both: 15+ uncaches `fetch` by default, and Cache Components (16) replaces route-segment caching with `'use cache'`.
1. Per piece of data on the page: is it per-request/per-user, or shared and cacheable? If cacheable, what staleness is acceptable (seconds, minutes, hours)?
2. Which UI elements genuinely need interactivity, state, effects, or browser APIs.
3. Mutations on the page and what data they invalidate.no-show-reduction
When your agent uses it
Cuts appointment no-shows for a local service business with a three-step reminder ladder (booking confirmation, day-before SMS, morning-of SMS with the tech's name and photo), confirmation-required policies, deposit rules for high-value appointments, and a reschedule-not-cancel script. Use when an owner says "customers keep not being home when the tech shows up", "how do I reduce no-shows", "should I take deposits for appointments", or "people cancel same-day and it kills my schedule". Do NOT use for chasing quotes after the visit happened - use quote-follow-up-sequences instead. Do NOT use for winning back lapsed customers - use win-back-campaign instead.
What it loads into context
Frontmatter
146 tokensAlways available to the agent
SKILL.md
2,224 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to elicit
- Concrete thresholds
- Worked artifact: Summit's no-show economics
- Deliverable
- Do NOT
- Quality bar
- Escalation and routing
Files
SKILL.mdcore9,388 bytes
Read SKILL.md
# No-Show Reduction
A no-show is the most expensive kind of nothing: the lead was paid for, the
slot was reserved, the tech drove out or sat idle, and no revenue arrived.
The costly mistake this skill prevents is treating no-shows as weather -
random and unmanageable - when a reminder ladder and a confirmation policy
reliably cut them by half or more. This skill and quote-follow-up-sequences
are the pack's two revenue-protection systems: both defend money the lead
spend already paid for.
Worked business throughout: Summit Plumbing, 4 techs, $450 average ticket,
$75,000 a month, booking roughly 300 appointments monthly. At its current 14
percent no-show rate that is about 42 missed visits; at the 65 percent close
rate that is roughly 14 lost jobs and $6,100 a month in gross revenue - thenonprofit-board-pack
When your agent uses it
Builds the nonprofit board meeting pack that produces engagement instead of nodding - a one-page dashboard of finances, program KPIs, and fundraising pipeline up front, routine items batched into a consent agenda, one real decision per meeting, and an explicit ask naming what board members can do. Use when a nonprofit ED says "prepare the board packet", "our board meetings are just report-outs", "the board approves everything without discussion", "what should go in the board dashboard", or "how do I get the board to actually help". Do NOT use for general board governance strategy, recruitment, or managing board dynamics - use board-management instead - and do NOT use for the public annual report - use impact-report-builder instead.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
2,256 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- Worked artifact: pack template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore10,050 bytes
Read SKILL.md
# Nonprofit Board Pack
Most nonprofit board meetings are serial report-outs: the ED talks for fifty minutes, the board nods, and the organization's best-connected volunteers go home having decided nothing and been asked for nothing. The board pack determines the meeting - a pack built as reports produces nodding; a pack built around a dashboard, a decision, and an ask produces a working board. This skill builds that pack. For governance strategy beyond the meeting - recruitment, composition, officer roles, managing a difficult chair - route to board-management; this skill owns the packet and the meeting it shapes.
Work the example organization throughout: Bright Futures Tutoring, a youth-education nonprofit with a $420,000 annual budget, 3 staff, serving 250 students, with a $180,000 grant goal and a $75,000 Hearthstone Foundation ask in play.
## Operating procedure
Send the pack 5-7 days before the meeting - late enough to be current, early enough that "I didn't have time to read it" is a choice. Under 10 pages total; the dashboard does the compression.
### Step 1: Build the one-page dashboard (the dashboard-first rule)
Page one is always a single-page dashboard with three blocks - finances, program KPIs, fundraising pipeline - each showing actual vs plan and direction vs last meeting. The rule exists because board members are part-time: they will reliably read one page, and if that page holds the whole organization's vitals, every member arrives oriented even if they read nothing else.
notion-database
When your agent uses it
Designs Notion databases that scale - one entity per database, deliberate property types, two-way relations with rollups, audience-specific views, and row templates - with explicit decision rules for relation vs rollup vs formula. Use when someone asks "design a Notion database for my projects", "should this be a relation or a rollup", "my Notion workspace is a junk drawer", or "set up a tasks and projects system in Notion". Do NOT use for building a full personal knowledge-management method with capture and review habits - use second-brain instead. For SQL or application database schemas, use database-schema.
What it loads into context
Frontmatter
144 tokensAlways available to the agent
SKILL.md
1,821 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: a three-database schema
- Deliverable
- Quality bar
- Do NOT
- Related skills
Files
SKILL.mdcore8,161 bytes
Read SKILL.md
# Notion Database Designer
A well-modeled Notion database scales; a sloppy one becomes a junk drawer where the same project name is typed forty different ways and no filter works. The expensive mistake is deferred: free-text fields and duplicated names feel fine for a month, then the workspace crosses a few hundred rows and every view, count, and cleanup becomes manual archaeology.
## Inputs to collect
1. What a single row represents, in one sentence (a task, a contact, a project). If a row could mean two different things, that is two databases - settle this first.
2. The questions people will ask of the data ("what's overdue?", "how loaded is each project?") - each question becomes a view or a rollup.
3. Who uses it and how often - each distinct audience gets its own view.
4. What already exists - related databases to link to rather than duplicate.
## Operating procedure
Order matters: entity before properties (properties describe the entity), relations before rollups (rollups aggregate across relations), structure before views (views are lenses on structure).nutrition-planner
When your agent uses it
Sets personal nutrition targets - protein at 1.6-2.2 g/kg, a calorie budget from goal and activity, fiber and hydration floors - and a repeatable meal structure with a review-and-adjust protocol for fat loss, muscle gain, or maintenance. Use when someone asks "how much protein should I eat", "set my macros for a cut", "why has my weight stalled", or "build me a sustainable diet framework". General wellness guidance, not medical advice. Do NOT use to produce a specific week of recipes with a shopping list and prep schedule - use meal-planner instead (this skill sets the targets; meal-planner cooks to them); for the broader multi-pillar health routine, use longevity-protocol.
What it loads into context
Frontmatter
165 tokensAlways available to the agent
SKILL.md
1,642 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,313 bytes
Read SKILL.md
# Nutrition Planner
Most diets fail on adherence, not knowledge: targets are set too aggressively, protein is an afterthought, and there is no rule for when and how to adjust. This skill sets defensible targets, wraps them in a repeatable meal structure, and installs a review protocol - so the plan survives month three, not just week one. It sets the numbers; turning them into a specific week of recipes, a shopping list, and a prep schedule is meal-planner's job.
## Operating procedure
Order matters: protein is set before calories because it is the highest-leverage macro and it anchors satiety; structure comes last because it exists to deliver the targets, not the reverse.
### Step 1: Gather inputs
Collect: body weight (kg or lbs), goal (fat loss, muscle gain, or maintenance), training days per week and type, occupation activity (desk vs on-feet), dietary pattern or restrictions (vegetarian, allergies - allergies are non-negotiable), rough current eating pattern, and past diet attempts with why they broke down. If weight or activity is a guess, label it a guess and use the conservative end of every range. Screen for the medical conditions in Escalation before setting any deficit.
### Step 2: Set the protein target
oauth-flow
When your agent uses it
Implements OAuth 2.0 and OpenID Connect correctly - authorization code with PKCE, client credentials for machine-to-machine, state and nonce, exact redirect URI matching, safe token storage, refresh rotation with reuse detection, scopes, ID token validation via JWKS, and logout - aligned with RFC 9700 (OAuth Security BCP) and RFC 7636 (PKCE). Use when someone asks "add Google or Microsoft login", "implement OAuth in our app", "where should I store the refresh token", "how do I validate an ID token", or is wiring an app to an identity provider or third-party API. Do NOT use for auditing a whole application's security posture - use security-audit instead; for line-level review of an auth diff, use secure-code-review.
What it loads into context
Frontmatter
177 tokensAlways available to the agent
SKILL.md
2,356 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Provider gotchas
- Thresholds
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore10,032 bytes
Read SKILL.md
# OAuth & OIDC Flow
OAuth bugs fail silently: a login that works in the demo can still accept forged callbacks, leak refresh tokens to page scripts, or link an attacker's account to a victim's. The costly mistake this skill prevents is copying a flow that "works" - implicit grant, tokens in localStorage, unchecked state, an ID token decoded but never verified - and shipping an account-takeover path no functional test will catch.
## Operating procedure
Decide the grant and token storage before writing a route; most OAuth flaws are architecture choices.
### Step 1: Choose the grant
- A user signs in (web, SPA, native): authorization code + PKCE. RFC 9700 requires PKCE for public clients and recommends it for confidential ones - use it for both.
- No user present: client credentials, confidential clients only.
- Never implicit (fragment tokens leak via history and scripts) or password credentials (RFC 9700: must not be used).
- SPAs: prefer a backend-for-frontend (BFF) that holds tokens and gives the browser only an httpOnly session cookie.objection-handler
When your agent uses it
Diagnoses what is actually behind a B2B sales objection - price, timing, authority, need, competitor, or status quo - and supplies a calibrated, non-pushy response for each with reframe patterns and exact language. Use when someone asks "they said it's too expensive, what do I say", "how do I handle 'not right now'", "the buyer says they're fine with their current tool", or a rep needs a response framework mid-deal. Do NOT use for gym or local-fitness lead objections and speed-to-lead cadence - use objection-handling-and-speed-to-lead instead; for writing a full call script, use closer-sales-script.
What it loads into context
Frontmatter
152 tokensAlways available to the agent
SKILL.md
1,627 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked pairs - one per category
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,582 bytes
Read SKILL.md
# Objection Handler
An objection is not a rejection - it is information. The costly mistake this prevents is the reflex rebuttal: answering the surface words ("too expensive") instead of the real concern underneath (unclear value, no budget authority, or a polite stall), which burns trust and teaches the buyer to stop sharing objections at all. Diagnose first, respond second, one objection at a time.
## Operating procedure
### Step 1: Gather inputs
Before crafting a response, establish:
1. The objection verbatim - the buyer's exact words, not the rep's paraphrase.
2. Deal stage (discovery, evaluation, negotiation) - the same words mean different things at different stages; "too expensive" in discovery is usually unclear value, in negotiation it is often real.
3. Who raised it and whether they hold budget authority.
4. What quantified pain has already been established. If none, that gap - not the objection - is the real problem.objection-handling-and-speed-to-lead
When your agent uses it
Builds the speed-to-lead system, multi-touch follow-up cadence, and four-objection response framework that stop a gym from losing leads to slow contact or stalled consults. Use when a gym owner or coach says "my leads go cold before we ever talk to them", "prospects stall at the price and never sign", "we are slow and inconsistent at follow-up", or a consult stalls on price, time, "ask my spouse", or "let me think about it" - and when building a follow-up sequence, no-show rebooking, or drilling objection responses with the team. Do NOT use when the buyer is a B2B account and the objection is an incumbent competitor or organizational status quo - use objection-handler instead.
What it loads into context
Frontmatter
162 tokensAlways available to the agent
SKILL.md
2,425 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Do NOT
- Calculator
- Follow-up sequence (scripts per state)
- Objection cards (quick reference for the floor)
Files
SKILL.mdcore9,765 bytesSKILL.mdcore9,765 bytesSKILL.mdcore9,765 bytes
Read SKILL.md
# Objection Handling and Speed to Lead
Two things lose gym leads: contacting them too slowly and folding at the first objection. This skill installs the speed-to-lead system, the multi-touch cadence, and the four-objection framework that fix both, scoped to gym and consumer-buyer prospects.
## Workflow
### Step 1: Contact every lead within minutes
Response time is the biggest lever in lead conversion. A lead is hottest the second they submit; minutes later they are browsing competitors.
1. Target first contact within 5 minutes during business hours, every time.
2. First touch is a call. No answer → text, then email - all three inside the first few minutes.
3. Assign an owner per lead with an alert or rotation so no lead waits. If the owner cannot answer, it routes to someone who can.
4. Run the calculator below to quantify what current response time costs in lost members, then set the 5-minute target against that number.
### Step 2: Run the multi-touch cadence by lead stateobservability-stack
When your agent uses it
Instruments systems with the three pillars - structured logs, RED/USE metrics, and distributed traces - correlated by one trace ID, with cardinality budgets and symptom-based alerts. Use when someone asks "how should I instrument this service", "what metrics should I emit", "why is my metrics bill exploding", "set up OpenTelemetry for my stack", or "we can't tell which service is slow". Do NOT use for tuning an existing noisy pager or alert thresholds - use alert-tuning instead; for defining SLO error-budget policy, use error-budget-policy; for the incident process when alerts fire, use sev-triage and postmortem-writer.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,758 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- The three pillars and what each is for
- Operating procedure
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,575 bytes
Read SKILL.md
# Observability Stack
Instrumentation exists to answer one question during an incident: what is wrong and why, from telemetry alone, without SSH-ing into boxes. The two expensive failure modes are opposite: under-instrumented systems where an outage is archaeology, and over-instrumented ones where a single high-cardinality label multiplies the metrics bill by 10,000x. This skill installs the three pillars with correlation between them and explicit cardinality budgets.
## The three pillars and what each is for
- Logs: discrete events with context - for debugging a specific request. Expensive per byte; highest detail.
- Metrics: pre-aggregated numbers over time - for dashboards and alerts. Cheap to query; no per-request detail.
- Traces: the path of one request across services - for locating which hop added the latency or threw the error.
The pillars are useless in isolation; the connective tissue is one correlation ID. A `trace_id` in every log line and exemplars on metrics let you pivot: alert fires (metric) → exemplar trace shows the slow hop (trace) → logs for that trace_id show the error (log). Instrument on OpenTelemetry so all three stay vendor-neutral.
## Operating procedure
offer-rejection-writer
When your agent uses it
Drafts unambiguous candidate offer letters and respectful, legally-careful candidate rejection notes for a human to review and send. Use when you are asked to write, draft, or word an offer letter, an offer email, or a candidate rejection/decline note - or to soften, tighten, or de-risk an existing draft of one. Do NOT use when writing performance feedback for a current employee - use feedback-writer instead.
What it loads into context
Frontmatter
109 tokensAlways available to the agent
SKILL.md
916 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Deliverable
- Do NOT
Files
SKILL.mdcore4,384 bytes
Read SKILL.md
Draft external candidate offer and rejection communications that are clear, warm, and legally careful; you draft, a human recruiter or hiring manager reviews and sends.
## Workflow
1. Identify which artifact you are drafting: an offer (extend) or a rejection (decline), and the candidate's stage (early-funnel screen vs. late-stage interviewed). Stage sets length, channel, and personalization.
2. Gather the inputs you must not invent. For offers: role title, level, base compensation, variable/equity with vesting basics, start date, location/remote terms, reporting line, response deadline - all from the comp/role owner. If any figure is missing, leave a clearly marked placeholder and flag it; never guess a number.
3. For an OFFER, draft in this order: specific enthusiasm tied to the interview → role title and level → base comp → variable/equity with vesting basics → start date → location/remote terms → reporting line → response deadline → a note that the letter is contingent on standard conditions and is not the full legal contract → a low-friction way to ask questions. A reasonable response window is 5-7 business days; anything under 72 hours reads as a pressure tactic and should be flagged back to the hiring team.
4. For a REJECTION, draft in this order: thank them for the specific time invested → deliver the decision plainly in the first two sentences → a truthful, job-related, non-comparative reason OR an honest statement that you cannot give individualized feedback → an open door only if it is genuinely real. Never bury the no.
5. Match effort to investment. A candidate who completed a screen gets a short, prompt note; one who invested 3+ hours of interviews (onsite, take-home, panel) warrants a personalized note or a call. Rejections should go out within 2-3 business days of the decision - every day of silence past a week actively damages the employer brand.
6. Check the legal and fairness bounds (see below) before finishing.
7. Hand back the draft marked as draft-for-review, and surface anything needing legal or comp sign-off.
## Quality bar
- An offer leaves zero ambiguity about terms: a reader knows exactly what they are being offered and by when to respond.
- Enthusiasm is specific to this candidate and this interview, not perks boilerplate.
- A rejection delivers the decision early, keeps the person's dignity intact, and creates no legal exposure.okr-builder
When your agent uses it
Drafts a quarter's OKRs - an inspiring qualitative objective with 3-5 measurable key results, each with a baseline, target, and owner - plus the weekly confidence check-in and end-of-quarter grading cadence. Use when someone asks "help me write OKRs", "are these key results any good", "turn our goals into OKRs for next quarter", or "how do we score our OKRs". Do NOT use for building the full-year company plan that OKRs slot into - use annual-plan instead - or for personal goal-tracking with a check-in partner - use goals-accountability instead.
What it loads into context
Frontmatter
141 tokensAlways available to the agent
SKILL.md
1,626 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example: bad/good pair
- Template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,551 bytes
Read SKILL.md
# OKR Builder
OKRs align effort around outcomes, and they fail in one predictable way: key results that are really a to-do list in disguise, so the team "completes" everything and moves nothing. The job of this skill is to produce few, measurable, honest OKRs - where a 0.7 at quarter end is a good result on a stretch goal, not a failure to explain away.
## Operating procedure
### Step 1: gather inputs
Collect before drafting; label anything the user estimates as a guess.
- The company or org priority this quarter rolls up to (one sentence). If the user cannot state it, that is the first problem - OKRs written in a vacuum get orphaned. If no annual plan exists, route to annual-plan first.
- The team's candidate goals, in whatever raw form exists.
- Current baselines for any metric that might become a key result. No baseline means no KR yet - default to spending the first two weeks of the quarter instrumenting it, and say so in the OKR.
- Who is available to own each key result.onboarding-copy
When your agent uses it
Writes product onboarding microcopy - welcome emails, empty states, tooltips, setup checklists, and success messages - mapped to the activation milestone so new users reach the aha moment with less friction. Use when someone asks "write our welcome email", "what should this empty state say", "improve our onboarding flow copy", or reports drop-off between signup and first value. Do NOT use for employee onboarding plans - use onboarding-plan-builder instead; for marketing landing pages, use landing-page-copy.
What it loads into context
Frontmatter
116 tokensAlways available to the agent
SKILL.md
755 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Process
- Touchpoints and how to write them
- Writing rules
- Tone
- Measuring success
- Anti-patterns
- Output
Files
SKILL.mdcore3,407 bytes
Read SKILL.md
# Onboarding Copy
You write the words a new user reads in their first session. The goal is activation: getting them to the moment they first feel the product's value (the "aha moment") with as little friction as possible.
## Process
1. Identify the activation milestone - the action that predicts a user will stick (e.g., "created first project," "invited a teammate").
2. Map the onboarding journey from signup to that milestone.
3. Write copy for each touchpoint that removes friction and builds momentum.
## Touchpoints and how to write them
### Welcome email
- Subject: warm, specific. Set the one next step.onboarding-plan-builder
When your agent uses it
Produces a 30/60/90-day onboarding plan for a specific role with phased milestones, first-week setup, observable success criteria, and ramp reviews. Use when a new hire has an accepted offer or start date, or when asked to write, draft, or improve a 30/60/90 / ramp / onboarding plan for a role. Do NOT use for the structure of ongoing recurring 1:1s - use 1on1-agenda instead.
What it loads into context
Frontmatter
114 tokensAlways available to the agent
SKILL.md
862 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Quality bar
- Deliverable
- Do NOT
Files
SKILL.mdcore3,783 bytes
Read SKILL.md
# Onboarding Plan Builder
Turn "get up to speed" into a sequenced, role-specific 30/60/90-day plan with concrete milestones and observable success criteria.
## Workflow
1. Capture the inputs. Pull the role's actual outcomes, level, team, manager, and start date. If the role's outcomes are unstated, ask for them or derive them from the job description before building - never plan against a generic title.
2. Phase the 90 days by cognitive load: Learn (0-30), Contribute (31-60), Own (61-90). Day 30 targets understanding systems, people, and context; day 60 targets shipping scoped work with support; day 90 targets owning an area with autonomy.
3. Attach 2-4 milestones to each phase, drawn from the role's real outcomes (not a generic checklist). Order milestones so each depends only on what earlier phases delivered.
4. Specify the first week concretely: access and tools provisioned before day one, the named buddy/onboarding partner, who to meet and why, where the docs live, and one small first task that produces a real win.
5. Write an observable success criterion per milestone - what "on track" looks like in evidence ("merged first PR," "ran a customer call solo," "presented the area roadmap"), never a feeling.
6. Define the three ramp reviews (day 30, 60, 90). Each covers progress vs. milestones, blockers, support needed, and a two-way fit check that invites feedback on the onboarding itself. Reference that recurring 1:1s exist; do not design their agenda here.
7. List the relationships and context to build: key cross-functional partners, rituals to join, and the implicit norms a newcomer cannot infer.
8. Calibrate to role and level, and flag where accommodations, visa, or relocation timing may shift the schedule.
oncall-handoff
When your agent uses it
Produces a complete on-call shift handoff note - shift summary, open incidents, watch items with paging thresholds, silenced alerts, in-flight changes, and escalation contacts - checked against a completeness checklist. Use when someone asks "write my on-call handoff", "what should I tell the next on-call", "my shift ends in an hour, help me hand over", or "make a handoff template for our rotation". Do NOT use for writing the incident postmortem after resolution - use postmortem-writer instead. For live severity assessment during an incident, use sev-triage; for external customer-facing incident comms, use status-page-update.
What it loads into context
Frontmatter
156 tokensAlways available to the agent
SKILL.md
1,844 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: filled handoff template
- Deliverable
- Do NOT
- Quality bar
- Related skills
Files
SKILL.mdcore7,767 bytes
Read SKILL.md
# On-Call Handoff
A handoff note transfers your mental model, not just your ticket queue. The incoming engineer should be able to respond to the next alert without asking you anything - because at 3am, "ask the previous on-call" is not an option, and the most expensive handoff failure is the one-line "quiet shift, nothing to report" that omits the silenced alert or the half-finished migration that pages them four hours later.
## Operating procedure
### Step 1: Gather the shift's raw material
Pull these before writing - memory alone misses exactly the items that matter:
1. Paging history for the shift: count and severity of pages.
2. Open or partially resolved incident tickets.
3. **Every alert you silenced, snoozed, or downgraded** - from the monitoring system, not memory. A silenced alert with no expiry is a landmine.
4. All deploys, config changes, migrations, and infra changes in the last 24 hours - including other teams' changes to shared systems if you know of them.one-pager-designer
When your agent uses it
Writes a decision-driving one-pager - problem, evidence, impact, and a specific ask compressed into a single page readable in under three minutes. Use when someone asks "turn this into a one-pager", "write a brief for leadership", "I need a one-page pitch for this decision", or is preparing a document a busy reader must act on from one page. Do NOT use for a client-facing sales proposal with pricing and scope - use sales-proposal-writer instead; for condensing an existing long document use executive-summary; for multi-slide narratives use slide-deck-builder.
What it loads into context
Frontmatter
139 tokensAlways available to the agent
SKILL.md
1,394 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked skeleton (copy and fill)
- Formatting rules
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore6,030 bytes
Read SKILL.md
# One-Pager Designer
A one-pager is a commitment device: confining the argument to one page forces the author to know what matters and discard what does not. It is not a short document; it is a complete argument at compression. The costly failure it prevents is the vague ask - a page that makes the reader feel informed but gives them nothing to say yes or no to, which means the meeting happens anyway and the page changed nothing.
A one-pager is not an executive summary. An executive summary condenses a longer document; a one-pager is the whole argument - it stands alone, requires no parent document, and is designed to be forwarded, printed, or shared independently.
## Inputs to collect
1. **The exact decision or action sought** - if the requester cannot state it in one sentence, extract it before writing anything else; the ask determines everything upstream of it.
2. **The reader** - who decides, what they already know, what they care about. Default to "knows the domain, not the details."
3. **The strongest fact or number** grounding the problem. If none exists, say so and label any estimate as an estimate.
4. **Three to five pieces of evidence** - facts or precedents, not assertions.
5. **The deadline or decision window**, if any.
op-ed-writer
When your agent uses it
Writes 600-800 word opinion pieces with a single contestable thesis, a timely news peg, evidence-backed argument paragraphs, an honestly-engaged counterargument, and a memorable closing call to action - flagging factual claims that need verification. Use when someone asks "write an op-ed about this issue", "help me pitch a piece to a newspaper", "my opinion piece feels wishy-washy", or wants to take a public stance on a current debate. Do NOT use for academic argumentation with formal citations - use academic-essay instead; for briefing policymakers rather than persuading the public, use policy-brief.
What it loads into context
Frontmatter
145 tokensAlways available to the agent
SKILL.md
829 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- What makes an op-ed work
- Process
- Structure
- Writing rules
- Voice and credibility
- Anti-patterns
- Output
Files
SKILL.mdcore3,643 bytesSKILL.mdcore3,643 bytes
Read SKILL.md
# Op-Ed Writer
You write op-eds: short, argument-driven opinion pieces (usually 600-800 words) that take a clear stance on a timely issue and try to move the reader.
## What makes an op-ed work
- **A single, clear argument.** One thesis, sharply stated. Op-eds fail when they're wishy-washy or try to argue three things.
- **Timeliness ("the peg").** Tie to a current event, anniversary, or debate so the reader knows why now.
- **A distinct voice and standing.** Why is THIS writer credible on THIS topic? Surface the author's experience or authority early.
- **Brevity.** Every sentence earns its place.
## Process
1. Gather: the issue, the writer's position, their relevant credibility, the news peg, and the change they want.open-house-follow-up
When your agent uses it
Converts open-house visitors into clients with door capture, a same-evening first touch, a 3-touch follow-up sequence, and buyer-signal triage by financing and timeline. Use when an agent says "follow up with my open house leads", "what do I send open house visitors", "my sign-in sheet goes nowhere", or "how do I work an open house". Do NOT use for generic B2B follow-up cadences - use sales-followup-cadence instead.
What it loads into context
Frontmatter
108 tokensAlways available to the agent
SKILL.md
2,050 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Inputs to collect
- The 3-touch sequence
- Buyer-signal triage
- Realistic conversion thresholds
- Deliverable
- Do NOT
- Quality bar
- Escalation and compliance
Files
SKILL.mdcore8,928 bytes
Read SKILL.md
# Open House Follow-Up
An open house without a follow-up system is unpaid weekend work: the visitors walk out, the sign-in sheet goes in a drawer, and the two real buyers in the room hire whichever agent contacts them first. The costly mistake this skill prevents is treating follow-up as optional or slow - the agent who touches a visitor the same evening wins the relationship before competitors even open the sheet.
Worked example throughout: Priya, a residential agent in a suburban market, 14 transactions last year at a $485,000 average sale price, building toward 24. Her Sunday open house at 742 Alder Court (the $499,000 listing from listing-description-writer) draws 11 visitor parties. Realistic yield: 2-3 parties are active buyers, 3-4 are 6-12 months out, the rest are neighbors and browsers. The gap between 14 and 24 transactions closes on the 6-12-month group - and that group is exactly who buyer-nurture-sequence exists for.
## Operating procedure
Order matters: capture quality determines everything downstream, and the same-evening touch decays in value by the hour.
1. **Set up capture at the door.** Digital sign-in (tablet or QR to a form) - paper sheets produce illegible phone numbers and no structure. Require name, mobile, email, and exactly ONE qualifying question: **"Where are you in your search?"** with four options: *actively looking / starting to look / just curious / I live nearby*. One question keeps sign-in friction near zero while splitting the room into workable segments. Greet every party and ask one conversational follow-up ("What are you comparing this one against?") - the answer goes in notes the moment they move on.
2. **Annotate within 30 minutes of closing the door.** For each party: search stage, anything learned about financing or timeline, what they reacted to in the house. Memory of 11 parties is gone by dinner.
3. **Send the same-evening first touch (Touch 1).** Non-negotiable. SMS, before 8 PM, personal and specific - template below. Same-evening beats next-morning because the visit is still emotionally live and because the other agents they met this weekend will not move until Monday.
4. **Run the rest of the 3-touch sequence** (day-2 email, day-5 call).outreach-sequence-designer
When your agent uses it
Use when you need to design the cadence that moves a cold prospect to a booked meeting across channels - the orchestration of touches, not the copy of any one message. Trigger phrases: "design an outreach sequence", "build a cold cadence", "how many touches should my sequence have", "multi-channel sequence", "what's a good follow-up cadence", "add LinkedIn and calls to my email sequence", "what's the spacing between touches". Chooses the channel mix, sets length and step spacing, defines follow-up logic, matches personalization to account tier, wires the "why now" signal, varies the angle across steps, and sets exit criteria. Do NOT use for writing a single email (use [[cold-email-craft]]), choosing who goes in the sequence (use [[prospect-list-builder]]), or building it inside a tool (use [[apollo-prospecting]]).
What it loads into context
Frontmatter
192 tokensAlways available to the agent
SKILL.md
2,399 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- When to use this skill
- The workflow
- Sequence template
- Worked example
- Deliverable
- Quality bar
- Common failure modes
Files
SKILL.mdcore10,147 bytes
Read SKILL.md
# Design a Multi-Channel Outreach Sequence
A sequence is a system, not a stack of emails. The unit of design is the *cadence*: which channels fire, in what order, how far apart, with what angle, and when a prospect exits. Most reps obsess over the words of touch #1 and ship a 3-email "sequence" that is really one email sent three times. That loses on both axes - single-channel, and no angle variation - so it gets ignored after the first send.
The core insight: replies are a function of *coverage over time*, not cleverness in a single message. The majority of positive replies land after the first touch, and a meaningful share come from channels other than email. Your job here is to design the orchestration so that a prospect who ignores your Day 1 email still gets a relevant, differently-angled reason to respond on Day 4 and Day 9 - without you babysitting it. The words live in [[cold-email-craft]]; the *structure* lives here.
## When to use this skill
- You are standing up a new outbound motion and need a defensible default cadence.
- Your current sequence is email-only and reply rates have flatlined.
- You are getting opens but no replies, and suspect the follow-up logic (not the copy) is the problem.
- You need to differentiate effort by account tier without building three separate machines.
- You have a fresh batch of [[buying-signal-tracker]] signals and need a "why now" cadence to act on them.
pagination-and-sync-engineer
When your agent uses it
Designs correct cursor pagination and incremental delta sync against a mutating dataset - cursor contracts, updated_at watermarks, delete propagation, checkpointing, and idempotent reprocessing. Use when someone says "my sync is skipping rows", "should I use cursor or offset pagination", "design the pagination contract for this list endpoint", "keep a local copy of this API in sync", or has an offset/page=N loop against changing data. Do NOT use for generating a typed API client or SDK from a spec - use api-client-generator instead; do NOT use for on-device offline-first sync with conflict resolution between a mobile client and server - use mobile-offline-sync instead; this skill owns server-to-server paging and one-way replication correctness.
What it loads into context
Frontmatter
168 tokensAlways available to the agent
SKILL.md
1,686 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: cursor pagination contract
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,365 bytesSKILL.mdcore7,365 bytesSKILL.mdcore7,365 bytesSKILL.mdcore7,365 bytesSKILL.mdcore7,365 bytesSKILL.mdcore7,365 bytes
Read SKILL.md
# Pagination and Sync Engineer
Make paging and delta sync correct against a dataset that mutates mid-walk: never skip, never double-count, and survive interruption. A naive page loop drops and duplicates rows the instant the underlying data changes, and the corruption is silent - you find out weeks later when a customer asks why a record is missing.
## Operating procedure
Steps 1-2 fix the read contract; 3-4 fix the sync semantics; 5-6 make it survivable. Order matters: a watermark on top of broken ordering is still broken.
### Step 1: Gather inputs
- Dataset size and write rate. A small, bounded, rarely-changing dataset (under ~10k rows, changes daily or less) can be fetched in one full pull - skip the machinery. Label estimates as estimates.
- Does the source expose a cursor, an updated_at filter, soft deletes or a tombstone/changes feed? Each missing capability forces a workaround below.
- Consistency target: how stale may the local copy be?
### Step 2: Choose cursor over offset - the decision rulepaid-acquisition-audit
When your agent uses it
Audits paid acquisition channels for wasted spend and untapped scaling headroom - computing break-even ROAS from gross margin, flagging creative fatigue and saturation, and producing a ranked action list with a dollar impact per item. Use when someone asks "where is my ad budget being wasted", "should I scale this campaign", "why did ROAS drop", or is preparing a budget review or planning to increase paid spend. Do NOT use for designing new creative experiments - use ad-creative-testing instead; for assigning cross-channel conversion credit, use marketing-attribution.
What it loads into context
Frontmatter
132 tokensAlways available to the agent
SKILL.md
1,765 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Audit checklist (ordered by spend impact)
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,656 bytes
Read SKILL.md
# Paid Acquisition Audit
A paid acquisition audit surfaces where money is bleeding and where the ceiling on profitable spend has not been reached yet. The costly mistake it prevents is acting on blended numbers: a "healthy" account average routinely hides campaigns burning cash below break-even next to starved winners that could absorb double the budget. The output is a ranked action list with a dollar impact per item - never a general report.
## Operating procedure
Work the steps in order. Break-even math (Step 2) must precede waste hunting (Step 3), because "waste" is undefined until the break-even line is drawn for this specific business.
### Step 1: Gather inputs
Collect before forming any opinion. Label any estimate a guess and revisit it in the findings.
1. Gross margin on the product sold through ads. This sets break-even ROAS. If unknown, default to 60% for DTC ecommerce or 80% for SaaS and label it a guess.
2. A 90-day export of spend, impressions, clicks, conversions, and revenue, broken down by channel, campaign, ad set, and ad. Never audit fewer than 30 days - short windows hide weekly patterns.pandas-expert
When your agent uses it
Writes correct, vectorized pandas for cleaning, joining, reshaping, and aggregating tabular data, with validated joins and deliberate dtype and missing-data handling. Use when someone asks "why did my merge duplicate rows", "how do I clean this CSV in pandas", "my groupby numbers look wrong", "this apply is too slow", or is transforming a DataFrame for analysis or a pipeline. Do NOT use for first-pass profiling of an unfamiliar dataset - use eda-playbook instead; for data too large for one machine use spark-jobs; for answering business questions directly in SQL use sql-to-insights.
What it loads into context
Frontmatter
143 tokensAlways available to the agent
SKILL.md
1,906 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked example
- Performance rules
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,929 bytesSKILL.mdcore7,929 bytesSKILL.mdcore7,929 bytesSKILL.mdcore7,929 bytesSKILL.mdcore7,929 bytes
Read SKILL.md
# Pandas Expert
Most pandas bugs are silent: a join that fans out rows, a dtype that upcast to object, a NaN that vanished from a sum. The output looks plausible and is wrong. This skill produces transformations that validate themselves - every join asserted, every coercion counted, every aggregate reconciled - so errors surface at the line that caused them instead of in a dashboard three weeks later.
## Operating procedure
Follow the steps in order. Types must be fixed before cleaning, cleaning before joining, joining before aggregating - each later step silently produces wrong numbers if an earlier one was skipped.
### Step 1: gather inputs
Before writing any transformation, establish:
- The grain of each table: what one row represents. If the user cannot state it, derive it with `df.duplicated(subset=candidate_keys).sum()` and label the result a guess until confirmed.
- The expected output: shape, grain, and one or two spot-check numbers the user already trusts (last month's total, a known customer's count).partition-planner
When your agent uses it
Produces a partitioning or sharding plan for an oversized table - range/list/hash strategy, partition-key choice, composite-key and pruning constraints, and a verdict on whether to partition at all. Use when one table has grown past tens of millions of rows or hundreds of GB, vacuum/autovacuum on it runs for hours, retention deletes are giant DELETEs, or you are deciding between range/list/hash partitioning or sharding a hot table. Do NOT use when the goal is choosing indexes for specific slow queries - use index-advisor instead.
What it loads into context
Frontmatter
128 tokensAlways available to the agent
SKILL.md
1,061 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Workflow
- Deliverable
- Quality bar
- Do NOT
Files
SKILL.mdcore5,069 bytesSKILL.mdcore5,069 bytes
Read SKILL.md
# Table Partition Planner
Plan how to split one too-large table into partitions (or shards) keyed to its real access and retention patterns - or prove it should stay a single indexed table.
## Workflow
1. **Confirm partitioning is warranted.** Get the row count and on-disk size. Below roughly 50-100M rows or a few hundred GB, a well-indexed single table usually beats a partitioned one; recommend staying single unless retention or vacuum pain forces the issue. Never partition to compensate for a missing index - that is index-advisor's job; add the index first and re-measure.
2. **Characterize the access pattern.** Capture the hot queries' WHERE clauses, the retention/deletion policy, and write distribution. The partition key must come from this evidence, not a guess.
3. **Pick the strategy from that pattern.**
- **Range** - time-series or any naturally ordered key (e.g. `created_at` by month). Old data leaves via `DROP`/`DETACH PARTITION` instead of a giant `DELETE`. Monthly granularity is the sane default for multi-year retention; drop to weekly or daily only when a single month would exceed roughly 50-100 GB.
- **List** - a small fixed set of discrete values (region, tenant tier).
- **Hash** - no natural range but you must cap any single child's size; use a fixed modulus (a power of two, commonly 8-64) to spread writes evenly, because changing the modulus later means rewriting the table.
4. **Choose the partition key.** It must appear in the WHERE clause of hot queries (or the planner cannot prune and scans every child), align with the retention policy (drop by time), and align with the largest indexes. Reject any column that is frequently updated - moving a row across partitions on UPDATE is expensive.
5. **Resolve the constraint and operational requirements.**
- In Postgres declarative partitioning, every unique constraint and primary key must include the partition key - design composite keys accordingly.partnership-strategy
When your agent uses it
Identifies, prioritizes, and structures business partnerships that produce measurable revenue - matching partnership type to goal, setting economics with real rev-share ranges, and enforcing a pilot-before-contract rule. Use when someone asks "should we partner with X", "how do we structure a reseller deal", "what rev share is normal", "our partnerships never produce anything", or "which partners should we prioritize". Do NOT use for choosing between partnerships and other distribution channels - use channel-strategy instead - or for the overall market-entry plan a partnership sits inside - use go-to-market-planner instead.
What it loads into context
Frontmatter
135 tokensAlways available to the agent
SKILL.md
1,576 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: deal one-pager template
- Deliverable
- Do NOT
- Quality bar
Files
SKILL.mdcore7,378 bytes
Read SKILL.md
# Partnership Strategy
Most partnerships are press releases that produce nothing: signed with fanfare, staffed by nobody, dead within two quarters. A real partnership creates measurable value for both sides and has a named owner accountable for a number. This skill separates strategic partnerships from logo-swap theater and structures the deal so the partner's people are actually incentivized to sell.
## Operating procedure
Fit-test before structuring, pilot before contract - in that order, because a well-structured deal with a bad-fit partner still produces nothing, and a deep contract before a proven motion locks in the failure.
### Step 1: Gather inputs
1. The goal - awareness, reach into a locked market, product stickiness, or a new revenue line. One primary goal; a partnership chasing three goals achieves none.
2. Candidate partners, if any exist, with what is known about their customer base.
3. Your ACV, margin, and sales motion - these bound what economics you can offer.
4. Capacity - who on your side would own this, at what fraction of their time. If the answer is nobody, stop here.patent-prior-art
When your agent uses it
Runs a systematic prior-art search - feature decomposition into claim elements, CPC/IPC classification plus keyword vocabulary, patent databases AND non-patent literature, forward and backward citation walks - and maps results into claim charts with a novelty assessment and a documented search log. Use when someone asks "is my invention novel", "has anyone already patented this", "find prior art against this claim", or is deciding whether a filing or an invalidity argument is worth pursuing. This is research, not legal advice; filing and freedom-to-operate decisions require a patent attorney. Do NOT use for a general academic literature survey unrelated to patentability - use literature-review instead.
What it loads into context
Frontmatter
150 tokensAlways available to the agent
SKILL.md
1,835 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Inputs to collect
- Operating procedure
- Worked artifact: search log template
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore8,555 bytes
Read SKILL.md
# Patent Prior Art
Assess whether an invention is novel and non-obvious by finding everything that already discloses it. The costly failure this skill prevents is the false-clear search: keywords-only, English-only, patents-only searching that misses the conference paper or Japanese utility model that later kills the application - after the filing fees, or worse, during litigation. This is a search-and-map task, not a legal opinion.
## Inputs to collect
1. **The invention disclosure**: what it does, how it works, and what the inventor believes is new.
2. **The priority date** (or intended filing date) - everything published on or after it is irrelevant, and date discipline is absolute.
3. **The field and its jargon**: how practitioners, not just patent drafters, name these concepts.
4. **Known competitors and prior products** in the space - likely assignees to search.
5. **Search depth**: quick patentability screen (top databases, hours) vs thorough pre-filing search (all sources below, days). Default to thorough if the user will spend money on the outcome.
## Operating procedure
performance-review-writer
When your agent uses it
Writes fair, specific, evidence-based performance reviews and ratings. Use when drafting or preparing for calibration to reduce recency bias and vague praise or criticism.
What it loads into context
Frontmatter
49 tokensAlways available to the agent
SKILL.md
832 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Before Writing: Gather Evidence First
- Rating Calibration
- Writing the Narrative
- Development Section
- Deliverable
- Quality bar
- Common Mistakes to Avoid
Files
SKILL.mdcore4,151 bytes
Read SKILL.md
# Performance Review Writer
Performance reviews are legally significant, career-defining, and often written in 20 minutes from memory. That gap causes recency bias, grade inflation, and feedback too vague to act on. This skill closes it.
## Before Writing: Gather Evidence First
Collect raw material from four sources before writing a word: the employee's self-review, peer feedback verbatims, the role's defined expectations for their level, and a concrete list of their work from the review period (PRs merged, projects owned, incidents handled, cross-functional contributions). Budget 2-3 hours per direct report for gathering and writing, not 20 minutes. Do not write from memory. If the list is hard to reconstruct, that is itself a signal about visibility and documentation.
## Rating Calibration
Start from the role expectations document, not a bell curve. Ask: did this person consistently meet, exceed, or fall short of the expectations defined for their level? If no level expectations document exists, write one before the review cycle, even a rough draft. Ratings applied without a standard are arbitrary. Sanity-check the distribution before submitting: if more than about a quarter of your team lands in the top rating band, recalibrate against the level expectations - either you have an exceptional team and can defend each case with evidence, or your bar has drifted.
## Writing the Narrative
pii-scrubber
When your agent uses it
Detects and redacts personally identifiable information from text, logs, datasets, and LLM prompts using layered pattern, checksum, and NER detection, then picks the right redaction mode for each downstream use. Use when someone asks "scrub PII from these logs", "redact personal data before sending it to the LLM", "is this dataset safe to share", or is wiring up logging, analytics, or third-party exports that might carry user data. Do NOT use for leaked API keys, tokens, or credentials - use secrets-hygiene instead.
What it loads into context
Frontmatter
134 tokensAlways available to the agent
SKILL.md
1,671 tokensWhen the skill is activated
Reference documents
0 tokensLoaded on request
Token counts are estimates. Scripts run separately; they are not counted as reference context.
The playbook
- Operating procedure
- Worked artifact: log line, bad vs good
- Compliance notes
- Edge cases
- Deliverable
- Do NOT
- Quality bar
- Escalation
Files
SKILL.mdcore7,450 bytes
Read SKILL.md
# PII Scrubber
Personal data that reaches logs, analytics, prompts, or exports is a liability that outlives the feature that leaked it - retention systems copy it, backups preserve it, and deletion requests cannot find it. The outcome of this skill is a scrubbing setup that catches PII at the boundary and redacts it in the least destructive mode that still protects the person. The costly mistake it prevents is redacting so naively that either PII slips through in free text or the data becomes useless for the job it was collected for.
## Operating procedure
Detection precedes mode selection because the right redaction depends on what the downstream consumer needs from each field - decide per field, not per dataset.
### Step 1: Gather inputs
- The data source: logs, database export, chat transcripts, prompt payloads, analytics events.
- Downstream use: debugging, analytics joins, support lookups, model training, third-party sharing. This decides the redaction mode.
- Whether records must remain joinable after scrubbing (tokenize) or only comparable for equality (hash) or neither (mask or remove).
- Jurisdictions and regimes in play (GDPR, CCPA, HIPAA). Default to the strictest plausible; label the assumption a guess.Skills included
- 1on1-agenda
- ab-test-analyzer
- abandoned-cart-sequence
- academic-essay
- accessibility-audit
- ad-creative-testing
- aeo-answer-blockifier
- agency-monthly-report
- agency-positioning
- agency-qbr-upsell
- agent-orchestration
- alert-tuning
- amazon-listing-optimizer
- animation-system
- annual-plan
- api-client-generator
- api-design
- api-versioning-strategist
- apollo-prospecting
- app-store-copy
- app-store-release-prep
- async-communication
- audience-targeting
- big-purchase-decision
- bloodwork-explainer
- board-management
- book-proposal
- brand-guidelines
- brand-naming
- budget-builder
- budget-pacing
- budget-vs-actual
- build-on-agent-native
- business-rule-extractor
- buyer-nurture-sequence
- buying-signal-tracker
- caching-strategy
- candidate-outreach-personalizer
- cap-table-manager
- captions-from-transcript
- carousel-scripter
- case-study-builder
- cash-flow-forecast
- category-page-copywriter
- causal-inference
- changelog-generator
- changelog-writer
- channel-strategy
- characterization-test-writer
- churn-reduction
- circuit-breaker-builder
- citation-tracker
- claims-verifier
- clickhouse-analytics
- client-churn-save
- client-comms-cadence
- client-onboarding-system
- client-status-reporting
- clinical-summary
- closer-sales-script
- cma-narrative-builder
- code-review-checklist
- cold-email-craft
- cold-email-deliverability
- color-accessibility
- color-palette-builder
- community-engagement-playbook
- comparison-page-builder
- competitive-intelligence
- competitive-moat
- component-api-design
- conflict-resolution
- connection-pool-tuner
- content-brief
- content-refresh-auditor
- content-repurposing
- contract-test-writer
- core-four-lead-engine
- course-outline-architect
- course-sales-page
- cover-letter-writer
- coverage-gap-finder
- creator-content-calendar
- crisis-comms-external
- cross-platform-reformatter
- csat-root-cause
- curriculum-mapper
- customer-analytics
- customer-success-qbr
- dark-mode-design
- data-drift-monitor
- data-quality
- data-room-builder
- data-story
- data-table-design
- database-schema
- dead-code-eliminator
- debt-payoff-planner
- decision-making
- deep-research
- deep-work-planner
- demo-script
- dependency-risk-audit
- design-critique
- design-handoff-doc
- design-qa-checklist
- design-token-system
- differentiated-instruction
- discovery-call-prep
- docker-compose-wizard
- document-template-system
- donor-communications
- due-diligence-checklist
- e2e-scenario-author
- economic-analysis
- eda-playbook
- email-deliverability
- email-drip-builder
- email-newsletter-pro
- email-triage
- emergency-fund-planner
- employment-contract
- eng-status-rollup
- engagement-reply-drafter
- env-doctor
- error-budget-policy
- error-handling
- error-message-writer
- escalation-summary
- executive-summary
- expansion-revenue
- expense-policy
- experiment-tracking
- expert-interview
- explain-plan-reader
- fact-checker
- feature-flags
- feature-store-design
- feedback-writer
- fiction-scene-writer
- finance-fx
- financial-planner
- financial-statement-builder
- fitness-program
- flaky-test-detangler
- flutter-widget-architect
- formatted-report-writer
- fpa-model
- framework-upgrader
- freelance-positioning
- fun-content
- fundraise-closing-mechanics
- fundraise-readiness-audit
- fundraise-team-hiring
- fundraising-narrative
- fundraising-stage-selector
- funnel-analysis
- geo-places
- ghostwriter
- git-commit-writer
- github-actions
- go-to-market-planner
- goals-accountability
- google-business-profile-optimizer
- government-open-data
- grand-slam-offer-builder
- grant-loi-writer
- grant-proposal-writer
- grant-prospect-researcher
- graphql-schema
- growth-accounting
- growth-model
- gtd-system
- gym-meta-ads-funnel
- gym-money-model
- gym-pricing-and-guarantees
- gym-transformation-challenge
- habit-builder
- hashtag-strategist
- help-documentation
- hiring-a-players-and-sops
- hiring-pipeline
- hiring-scorecard
- icon-system
- icp-persona-builder
- idempotency-enforcer
- image-license-rights
- impact-report-builder
- inbox-zero
- index-advisor
- injury-prehab
- instagram-post-builder
- instagram-trend-scout
- internal-linking-mapper
- interview-debrief-synthesizer
- interview-guide-builder
- interview-question-kit
- interview-synthesis
- investment-basics
- investor-pipeline-crm
- investor-targeting
- investor-update-writer
- jd-bias-scrubber
- jetpack-compose-builder
- jira-ticket-writer
- job-application
- job-description-writer
- job-interview-prep
- journal-framework
- jtbd-extractor
- kafka-pipelines
- kb-article-writer
- keyword-cluster-builder
- kinetic-typography
- kpi-scoreboard-and-cadence
- kubernetes-basics
- landing-page-copy
- landing-page-cro
- language-learning
- language-reference
- language-version-migrator
- launch-day-runbook
- launch-email-sequence
- launch-plan-sequencer
- lead-enrichment
- lesson-plan-builder
- life-coach
- lifecycle-journey-map
- linear-workflow
- linkedin-post-writer
- listing-description-writer
- literature-review
- llm-evaluation
- load-testing
- local-referral-engine
- local-seo-playbook
- local-unit-economics
- localization-guide
- logo-brief-writer
- longevity-protocol
- m-and-a-analysis
- manuscript-reviser
- market-sizing
- marketing-attribution
- mcp-builder
- meal-planner
- media-monitor
- meeting-agenda
- meeting-notes-to-actions
- messaging-hierarchy
- meta-title-optimizer
- microservices
- migration-safety-checker
- ml-feature-engineering
- mobile-offline-sync
- mobile-perf-profiler
- mobility-routine
- mock-stub-designer
- model-card-writer
- model-evaluation-report
- mongodb-expert
- monolith-decomposer
- monorepo-expert
- month-end-close
- moodboard-builder
- motion-color-and-light
- motion-design-principles
- motion-spec
- mutation-test-runner
- mutual-action-plan
- n-plus-one-hunter
- narration-script
- networking-system
- next-on-vercel-perf
- nextjs-app-router
- no-show-reduction
- nonprofit-board-pack
- notion-database
- nutrition-planner
- oauth-flow
- objection-handler
- objection-handling-and-speed-to-lead
- observability-stack
- offer-rejection-writer
- okr-builder
- onboarding-copy
- onboarding-plan-builder
- oncall-handoff
- one-pager-designer
- op-ed-writer
- open-house-follow-up
- outreach-sequence-designer
- pagination-and-sync-engineer
- paid-acquisition-audit
- pandas-expert
- partition-planner
- partnership-strategy
- patent-prior-art
- performance-review-writer
- pii-scrubber
Showing 300 of 482 skills. See GitHub for the complete repository.
Repository facts
- License
- MIT
- Created
- Jul 5, 2026
- Last pushed
- Oct 13, 2026
- Forks
- 3
- Open issues
- 0
- Project activity
- Active
Before you install
Network or broad shellAutomated signals, not a guarantee
Checked Nov 6, 2026
Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.
Show 478 notes
skills/ab-test-analyzer/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/academic-essay/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/agency-qbr-upsell/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/animation-system/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/audience-targeting/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/bloodwork-explainer/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/build-on-agent-native/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/business-rule-extractor/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/business-rule-extractor/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/buyer-nurture-sequence/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/buying-signal-tracker/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/buying-signal-tracker/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/candidate-outreach-personalizer/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/causal-inference/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzedskills/causal-inference/SKILL.md:1Analysis incomplete: Referenced executable target unavailable; not analyzed
- No broad tool permissions
- 4 dangerous shell commandsin 2 files: env-doctor/SKILL.md:132 vercel-env-management/SKILL.md:113
- 156 network referencesin 32 files: test-placeholder-data/SKILL.md:33 python-async/SKILL.md:35 fun-content/SKILL.md:33 and 29 more
- No hidden or obfuscated text
- No instruction overrides
- No executable scripts
- 45 references to secretsin 9 files: vercel-ai-gateway/SKILL.md:3 vercel-env-management/SKILL.md:3 readme-generator/SKILL.md:31 and 6 more
- No requests to vote in Hot
Show all 205 signals by file
skills/api-client-generator/SKILL.md
- Network accessline 28
export class UserService { static async getUser(id: string) { const res = await fetch(`${BASE}/users/${id}`, { - Network accessline 29
static async getUser(id: string) { const res = await fetch(`${BASE}/users/${id}`, { signal: AbortSignal.timeout(5000), // added by hand - lost on regen - Network accessline 30
const res = await fetch(`${BASE}/users/${id}`, { - Network accessline 41
// transport.ts - the only file that knows about timeouts, auth, errors export const transport: Fetch = async (url, init) => { const res = await fetch(url, { - Network accessline 42
export const transport: Fetch = async (url, init) => { const res = await fetch(url, { ...init, - Network accessline 43
const res = await fetch(url, {
skills/citation-tracker/SKILL.md
- Network accessline 56
Bad: Smith, John. "Effects Of Sleep On Memory." Journal of Cognition, 2020, https://doi.org/10.1234/fake - full first name (APA uses initials), title case on the article, missing volume/pages, unverified DOI. - Network accessline 58
Good: Smith, J. (2020). Effects of sleep on memory consolidation in adults. *Journal of Cognition*, 12(3), 145-162. https://doi.org/10.1234/jcog.2020.0312 - initials, sentence case, italicized container, volume(issue) and pages present, DOI
skills/docker-compose-wizard/SKILL.md
- Network accessline 70
- "[FILL: host_port]:8080" # only the app is host-reachable healthcheck: test: ["CMD", "curl", "-f", "http://localhost:8080/health"] - Network accessline 71
healthcheck: test: ["CMD", "curl", "-f", "http://localhost:8080/health"] interval: 10s - Network accessline 72
test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
skills/env-doctor/SKILL.md
- Access to secretsline 53
[Credential-related source omitted] - Access to secretsline 54
[Credential-related source omitted] - Network accessline 112
$ nvm use -> Now using node v20.18.0 $ hash -r && node -v -> v20.18.0 $ npm run dev -> ready on http://localhost:3000 - Network accessline 114
$ npm run dev -> ready on http://localhost:3000 - Dangerous shell operationline 132
- Do not start with `rm -rf node_modules package-lock.json` - it hides the cause and upgrades transitive dependencies. - Dangerous shell operationline 134
- Do not `sudo npm install -g` or `sudo pip install` to fix permission errors - fix the prefix or use a version manager or venv.
skills/error-handling/SKILL.md
- Network accessline 84
const input = parseInput(req); // 400 on failure - terminal const order = await withRetry( // taxonomy-driven retry () => orders.fetch(input.orderId), - Network accessline 85
const order = await withRetry( // taxonomy-driven retry () => orders.fetch(input.orderId), { attempts: 3, baseMs: 200, capMs: 20_000, retryOn: (e) => e.retryable === true }, - Network accessline 86
() => orders.fetch(input.orderId),
skills/finance-fx/SKILL.md
- Network accessline 31
https://api.frankfurter.dev/v1/latest?base=USD&symbols=EUR,GBP,JPY - Network accessline 43
https://api.frankfurter.dev/v1/2024-01-15?base=USD&symbols=EUR ← one date - Network accessline 44
https://api.frankfurter.dev/v1/2024-01-02..2024-01-31?base=USD&symbols=EUR ← range - Network accessline 45
https://api.frankfurter.dev/v1/currencies ← code → name map - Network accessline 53
https://api.coingecko.com/api/v3/simple/price?ids=bitcoin,ethereum&vs_currencies=usd&include_24hr_change=true - Network accessline 65
https://api.coingecko.com/api/v3/coins/markets?vs_currency=usd&order=market_cap_desc&per_page=20&page=1
skills/fun-content/SKILL.md
- Network accessline 33
https://opentdb.com/api_token.php?command=request ← once per user session - Network accessline 34
[Credential-related source omitted] - Network accessline 35
[Credential-related source omitted] - Network accessline 36
https://opentdb.com/api_category.php ← id → name list - Network accessline 37
[Credential-related source omitted] - Network accessline 53
https://catfact.ninja/fact → {"fact":"...","length":94} - Network accessline 54
https://random.dog/woof.json → {"url":"https://random.dog/….jpg","fileSizeBytes":170059} - Network accessline 55
https://api.adviceslip.com/advice → {"slip":{"id":89,"advice":"Don't be afraid to ask questions."}} - Network accessline 56
https://api.adviceslip.com/advice/search/{q} → {"slips":[...]} - 404s (JSON body) on no match - Network accessline 57
https://uselessfacts.jsph.pl/api/v2/facts/random → {"text":"...","source":"...","permalink":"..."}
skills/geo-places/SKILL.md
- Network accessline 32
https://nominatim.openstreetmap.org/search?q={URL-encoded query}&format=jsonv2&limit=1 - Network accessline 46
https://nominatim.openstreetmap.org/reverse?lat={lat}&lon={lon}&format=jsonv2 - Network accessline 54
https://geocoding-api.open-meteo.com/v1/search?name={city}&count=1 - Network accessline 67
https://api.zippopotam.us/{country}/{code} ← e.g. /us/90210, /de/10115 - Network accessline 80
https://www.apicountries.com/name/{name} ← fuzzy, returns array - Network accessline 81
https://www.apicountries.com/alpha/{code} ← ISO alpha-2/alpha-3, returns object - Network accessline 85
[{"name":"Japan","capital":"Tokyo","region":"Asia","population":126476461, "currencies":[{"code":"JPY","symbol":"¥"}],"languages":[{"name":"Japanese"}], "alpha2Code":"JP","flags":{"png":"https://flagcdn.com/w320/jp.png"}}] - Network accessline 87
"alpha2Code":"JP","flags":{"png":"https://flagcdn.com/w320/jp.png"}}]
skills/government-open-data/SKILL.md
- Network accessline 19
1. **Resolve the indicator code first.** Verified ones to use directly: `NY.GDP.MKTP.CD` (GDP, current US$), `SP.POP.TOTL` (population), `SP.DYN.LE00.IN` (life expectancy), `FP.CPI.TOTL.ZG` (inflation %), `SL.UEM.TOTL.ZS` (unemployment %). - Network accessline 32
https://api.worldbank.org/v2/country/US/indicator/SP.POP.TOTL?format=json&mrv=10 - Network accessline 33
https://api.worldbank.org/v2/country/US;CN;DE/indicator/NY.GDP.MKTP.CD?format=json&mrv=1&per_page=100 - Network accessline 34
https://api.worldbank.org/v2/country/US/indicator/SP.POP.TOTL?format=json&date=2013:2022 ← explicit range - Network accessline 49
https://api.worldbank.org/v2/country/JPN?format=json - Network accessline 57
https://ec.europa.eu/eurostat/api/dissemination/statistics/1.0/data/{dataset}?format=JSON&geo=DE&time=2023
skills/image-license-rights/SKILL.md
- Network accessline 35
- CC0 / Public Domain - do anything, no attribution, no permission (many museum and government archives). Still check for identifiable people and property (Step 4). - Creative Commons variants - the letters are the obligations: BY requires - Network accessline 36
- Creative Commons variants - the letters are the obligations: BY requires exact attribution as specified; SA requires derivatives carry the same license; NC forbids commercial use entirely; ND forbids modification. CC BY-NC on a product pa
skills/language-reference/SKILL.md
- Network accessline 30
https://api.dictionaryapi.dev/api/v2/entries/en/{word} - Network accessline 34
[{"word":"serendipity","phonetic":"/ˌsɛ.ɹən.ˈdɪ.pɪ.ti/", "phonetics":[{"text":"/…/","audio":"https://api.dictionaryapi.dev/media/pronunciations/en/serendipity-us.mp3"}], - Network accessline 35
"phonetics":[{"text":"/…/","audio":"https://api.dictionaryapi.dev/media/pronunciations/en/serendipity-us.mp3"}], - Network accessline 46
https://en.wikipedia.org/api/rest_v1/page/summary/{Title_With_Underscores} - Network accessline 50
{"title":"Albert Einstein","description":"German-born theoretical physicist", "extract":"Albert Einstein was a German-born theoretical physicist…", "thumbnail":{"source":"https://upload.wikimedia.org/...jpg","width":244}, - Network accessline 51
"extract":"Albert Einstein was a German-born theoretical physicist…", "thumbnail":{"source":"https://upload.wikimedia.org/...jpg","width":244}, "content_urls":{"desktop":{"page":"https://en.wikipedia.org/wiki/Albert_Einstein"}}} - Network accessline 52
"thumbnail":{"source":"https://upload.wikimedia.org/...jpg","width":244}, - Network accessline 53
"content_urls":{"desktop":{"page":"https://en.wikipedia.org/wiki/Albert_Einstein"}}} - Network accessline 61
https://en.wikipedia.org/w/rest.php/v1/search/page?q={query}&limit=5
skills/load-testing/SKILL.md
- Network accessline 61
}; export default function () { const res = http.get(`https://api.example.com/items?u=${__VU}-${__ITER}`); - Network accessline 62
export default function () { const res = http.get(`https://api.example.com/items?u=${__VU}-${__ITER}`); check(res, { 'status 200': (r) => r.status === 200 }); - Network accessline 63
const res = http.get(`https://api.example.com/items?u=${__VU}-${__ITER}`);
skills/mcp-builder/SKILL.md
- Network accessline 58
- Act as an OAuth 2.1 resource server. Unauthenticated requests get `401` with `WWW-Authenticate: Bearer resource_metadata="https://host/.well-known/oauth-protected-resource"`.
skills/mock-stub-designer/SKILL.md
- Network accessline 80
repo = InMemoryOrderRepository() # fake, reused across the suite responses_mock.post( # wire-level stub: URL + JSON exercised "https://api.pay.example/v1/charges", - Network accessline 81
responses_mock.post( # wire-level stub: URL + JSON exercised "https://api.pay.example/v1/charges", json={"id": "ch_1", "status": "succeeded"}, status=200, - Network accessline 82
"https://api.pay.example/v1/charges",
skills/next-on-vercel-perf/SKILL.md
skills/oauth-flow/SKILL.md
skills/observability-stack/SKILL.md
- Network accessline 74
for: 5m annotations: runbook: https://runbooks.internal/checkout-errors - Network accessline 75
annotations: runbook: https://runbooks.internal/checkout-errors summary: ">2% of checkouts failing for 5m - users cannot pay" - Network accessline 76
runbook: https://runbooks.internal/checkout-errors
skills/python-async/SKILL.md
- Network accessline 35
async def main(): async with asyncio.TaskGroup() as tg: t1 = tg.create_task(fetch(url1)) - Network accessline 36
async with asyncio.TaskGroup() as tg: t1 = tg.create_task(fetch(url1)) t2 = tg.create_task(fetch(url2)) - Network accessline 37
t1 = tg.create_task(fetch(url1)) - Network accessline 38
t2 = tg.create_task(fetch(url2)) - Network accessline 74
results = [] for u in urls: # runs one at a time results.append(await fetch(u)) - Network accessline 75
for u in urls: # runs one at a time results.append(await fetch(u)) data = requests.get(META_URL) # freezes the entire event loop - Network accessline 76
results.append(await fetch(u)) - Network accessline 77
data = requests.get(META_URL) # freezes the entire event loop - Network accessline 99
async with sem: async with asyncio.timeout(5): return await fetch(client, u) - Network accessline 100
async with asyncio.timeout(5): return await fetch(client, u) async with asyncio.TaskGroup() as tg: - Network accessline 101
return await fetch(client, u)
skills/readme-generator/SKILL.md
- Access to secretsline 31
[Credential-related source omitted] - Network accessline 52
1. **Name and one-sentence description**: what it is and who it is for, in plain words, within the first three lines. 2. **Short status line or badges**, only if they are true and maintained (a CI badge for a workflow that exists, a package - Network accessline 53
2. **Short status line or badges**, only if they are true and maintained (a CI badge for a workflow that exists, a package version). Use three badges at most. 3. **Quickstart**: prerequisites with versions, then the shortest verified path t - Network accessline 54
3. **Quickstart**: prerequisites with versions, then the shortest verified path to a visible result, ending with how to know it worked ("open http://localhost:3000 and you should see the sign-in page"). - Access to secretsline 83
[Credential-related source omitted] - Network accessline 94
git clone https://github.com/acme/ledgerly && cd ledgerly - Access to secretsline 95
[Credential-related source omitted] - Network accessline 102
Open http://localhost:3000. You should see the sign-in page. - Access to secretsline 110
[Credential-related source omitted] - Access to secretsline 113
[Credential-related source omitted]
skills/remotion-compose/SKILL.md
- Network accessline 239
Save the finished file to `src/compositions/<VideoName>.tsx`, register it in `src/Root.tsx`, and tell the user to preview with `npx remotion studio` (http://localhost:3000). Then hand off: *"Scrub it in Studio. Tell me what to - Network accessline 240
`src/Root.tsx`, and tell the user to preview with `npx remotion studio` (http://localhost:3000). Then hand off: *"Scrub it in Studio. Tell me what to change - too fast, wrong color, different copy - and I'll edit and re-render - Network accessline 241
(http://localhost:3000). Then hand off: *"Scrub it in Studio. Tell me what to
skills/remotion-render/SKILL.md
- Network accessline 60
Keep Studio open (`npx remotion studio`, http://localhost:3000) so the user previews instantly; reserve full renders for sign-off.
skills/remotion-setup/SKILL.md
- Network accessline 175
This serves the Studio at **http://localhost:3000**. The `create-video` template
skills/runbook-writer/SKILL.md
- Network accessline 63
Good: "Run `kubectl -n payments exec deploy/api -- pg_isready -h db.payments.internal -p 5432`. Expect `accepting connections`. If it fails or times out, skip to Mitigation 3 (fail over to replica - requires IC approval) and page #db-oncall
skills/secrets-hygiene/SKILL.md
- Access to secretsline 44
[Credential-related source omitted]
skills/seo-optimizer/SKILL.md
skills/space-earth-science/SKILL.md
- Network accessline 31
https://earthquake.usgs.gov/fdsnws/event/1/query?format=geojson - Network accessline 38
{"features":[{"properties":{"mag":6.2,"place":"58 km W of Tobelo, Indonesia", "time":1783045888846,"tsunami":0,"alert":"green","felt":7, "url":"https://earthquake.usgs.gov/earthquakes/eventpage/us6000t9r5"}, - Network accessline 39
"time":1783045888846,"tsunami":0,"alert":"green","felt":7, "url":"https://earthquake.usgs.gov/earthquakes/eventpage/us6000t9r5"}, "geometry":{"coordinates":[127.63,1.73,10.0]}}], - Network accessline 40
"url":"https://earthquake.usgs.gov/earthquakes/eventpage/us6000t9r5"}, - Network accessline 50
[Credential-related source omitted] - Access to secretsline 50
[Credential-related source omitted] - Network accessline 54
{"date":"...","title":"...","explanation":"...","media_type":"image", "url":"https://apod.nasa.gov/apod/image/...jpg","hdurl":"..."} - Network accessline 55
"url":"https://apod.nasa.gov/apod/image/...jpg","hdurl":"..."} - Network accessline 63
[Credential-related source omitted] - Access to secretsline 63
[Credential-related source omitted] - Network accessline 71
https://epic.gsfc.nasa.gov/api/natural ← latest image list - Network accessline 72
https://epic.gsfc.nasa.gov/archive/natural/{yyyy}/{mm}/{dd}/png/{image}.png
skills/stock-photo-api/SKILL.md
- Access to secretsline 26
[Credential-related source omitted] - Access to secretsline 27
[Credential-related source omitted] - Access to secretsline 36
[Credential-related source omitted] - Access to secretsline 37
[Credential-related source omitted] - Network accessline 39
// 1. Search. Hotlink the urls Unsplash returns; do not rehost the file. async function searchPhotos(query: string) { const res = await fetch( - Network accessline 40
async function searchPhotos(query: string) { const res = await fetch( `https://api.unsplash.com/search/photos?query=${encodeURIComponent(query)}&per_page=12&orientation=landscape`, - Network accessline 41
const res = await fetch( - Network accessline 42
`https://api.unsplash.com/search/photos?query=${encodeURIComponent(query)}&per_page=12&orientation=landscape`, - Network accessline 52
// Skipping it is the top reason Unsplash deactivates apps. async function triggerDownload(downloadLocation: string) { await fetch(downloadLocation, { headers: H }); - Network accessline 53
async function triggerDownload(downloadLocation: string) { await fetch(downloadLocation, { headers: H }); } - Network accessline 54
await fetch(downloadLocation, { headers: H }); - Network accessline 59
const u = `${photo.user.links.html}?utm_source=your_app&utm_medium=referral`; return `Photo by <a href="${u}">${photo.user.name}</a> on ` + `<a href="https://unsplash.com/?utm_source=your_app&utm_medium=referral">Unsplash</a>`; - Network accessline 60
return `Photo by <a href="${u}">${photo.user.name}</a> on ` + `<a href="https://unsplash.com/?utm_source=your_app&utm_medium=referral">Unsplash</a>`; } - Network accessline 61
`<a href="https://unsplash.com/?utm_source=your_app&utm_medium=referral">Unsplash</a>`;
skills/test-placeholder-data/SKILL.md
- Network accessline 33
https://dummyjson.com/products/1 - Network accessline 34
https://dummyjson.com/products?limit=10&skip=20&select=title,price,rating - Network accessline 35
https://dummyjson.com/products/search?q=phone - Network accessline 36
https://dummyjson.com/users https://dummyjson.com/carts https://dummyjson.com/posts - Network accessline 50
https://jsonplaceholder.typicode.com/posts/1 - Network accessline 51
https://jsonplaceholder.typicode.com/posts?userId=1 - Network accessline 52
https://jsonplaceholder.typicode.com/posts/1/comments - Network accessline 64
https://randomuser.me/api/?results=25&nat=us,gb&seed=demo - Network accessline 68
{"results":[{"name":{"first":"Regina","last":"Fisher"},"email":"regina.fisher@example.com", "location":{"city":"York","state":"Massachusetts","country":"United States"}, "picture":{"large":"https://randomuser.me/api/portraits/women/88.j - Network accessline 69
"location":{"city":"York","state":"Massachusetts","country":"United States"}, "picture":{"large":"https://randomuser.me/api/portraits/women/88.jpg"}, "login":{"uuid":"c305bcf6-..."}}],"info":{"seed":"demo","results":25}} - Network accessline 70
"picture":{"large":"https://randomuser.me/api/portraits/women/88.jpg"}, - Network accessline 79
https://picsum.photos/id/237/400/300 ← deterministic (dog photo, 400×300) - Network accessline 80
https://picsum.photos/seed/{any-string}/400/300 ← deterministic by seed - Network accessline 81
https://picsum.photos/v2/list?page=1&limit=30 ← JSON metadata: id, author, download_url
skills/vercel-ai-gateway/SKILL.md
- Access to secretsline 3
[Credential-related source omitted] - Access to secretsline 42
[Credential-related source omitted] - Access to secretsline 50
[Credential-related source omitted] - Access to secretsline 54
[Credential-related source omitted] - Access to secretsline 55
[Credential-related source omitted] - Network accessline 150
# Live budget + spend for one AI Gateway key. curl "https://ai-gateway.vercel.sh/v1/quotas?quotaEntityId=api_key_id_<your_key_id>" \ - Network accessline 151
[Credential-related source omitted] - Access to secretsline 151
[Credential-related source omitted] - Access to secretsline 152
[Credential-related source omitted] - Access to secretsline 171
[Credential-related source omitted] - Access to secretsline 189
[Credential-related source omitted] - Access to secretsline 280
[Credential-related source omitted] - Access to secretsline 283
[Credential-related source omitted] - Access to secretsline 285
[Credential-related source omitted] - Network accessline 332
[Credential-related source omitted] - Access to secretsline 333
[Credential-related source omitted]
skills/vercel-deploy-pipeline/SKILL.md
- Access to secretsline 3
[Credential-related source omitted] - Network accessline 60
vercel inspect <preview-url> # status, build, aliases vercel curl /api/health --deployment <preview-url> # bypasses deploy protection - Network accessline 61
vercel curl /api/health --deployment <preview-url> # bypasses deploy protection - Network accessline 221
echo "==> Hitting $HEALTH (deploy-protection bypassed)" if ! vercel curl "$HEALTH" --deployment "$URL" >/dev/null 2>&1; then - Network accessline 222
if ! vercel curl "$HEALTH" --deployment "$URL" >/dev/null 2>&1; then - Network accessline 239
$ ./promote-gate.sh https://acme-7gk2qz.vercel.app - Network accessline 240
==> Inspecting https://acme-7gk2qz.vercel.app - Network accessline 242
==> Scanning error logs ==> PASS. Promote the SAME deployment (no rebuild) with: vercel promote https://acme-7gk2qz.vercel.app --yes && vercel promote status - Network accessline 244
vercel promote https://acme-7gk2qz.vercel.app --yes && vercel promote status - Access to secretsline 280
[Credential-related source omitted] - Access to secretsline 297
[Credential-related source omitted] - Access to secretsline 302
[Credential-related source omitted]
skills/vercel-edge-and-isr/SKILL.md
- Network accessline 103
// serves stale and regenerates in the background. export default async function Page() { const res = await fetch('https://api.example.com/posts', { - Network accessline 104
export default async function Page() { const res = await fetch('https://api.example.com/posts', { next: { revalidate: 60, tags: ['posts'] }, - Network accessline 105
const res = await fetch('https://api.example.com/posts', { - Network accessline 145
const res = await fetch('https://api.example.com/products'); - Network accessline 182
export default { async fetch(request: Request) { - Network accessline 183
async fetch(request: Request) { - Network accessline 189
const data = await (await fetch('https://api.example.com/blog')).json();
skills/vercel-env-management/SKILL.md
- Access to secretsline 3
[Credential-related source omitted] - Access to secretsline 89
[Credential-related source omitted] - Access to secretsline 96
[Credential-related source omitted] - Access to secretsline 97
[Credential-related source omitted] - Dangerous shell operationline 113
[Credential-related source omitted] - Access to secretsline 113
[Credential-related source omitted] - Dangerous shell operationline 114
[Credential-related source omitted] - Access to secretsline 114
[Credential-related source omitted] - Access to secretsline 135
[Credential-related source omitted] - Network accessline 166
audience: `//iam.googleapis.com/projects/${process.env.GCP_PROJECT_NUMBER}/locations/global/workloadIdentityPools/${process.env.GCP_WORKLOAD_IDENTITY_POOL_ID}/providers/${process.env.GCP_WORKLOAD_IDENTITY_POOL_PROVIDER_ID}`, subject_token - Network accessline 167
subject_token_type: 'urn:ietf:params:oauth:token-type:jwt', token_url: 'https://sts.googleapis.com/v1/token', service_account_impersonation_url: `https://iamcredentials.googleapis.com/v1/projects/-/serviceAccounts/${process.env.GCP_SERV - Network accessline 168
token_url: 'https://sts.googleapis.com/v1/token', - Network accessline 169
service_account_impersonation_url: `https://iamcredentials.googleapis.com/v1/projects/-/serviceAccounts/${process.env.GCP_SERVICE_ACCOUNT_EMAIL}:generateAccessToken`, - Access to secretsline 179
[Credential-related source omitted] - Access to secretsline 309
[Credential-related source omitted] - Access to secretsline 328
[Credential-related source omitted] - Access to secretsline 338
[Credential-related source omitted]
skills/weather-climate/SKILL.md
- Network accessline 19
1. **Get coordinates first.** Every endpoint below takes `latitude`/`longitude`, not place names. If you only have a city name, resolve it with Open-Meteo's geocoder: `https://geocoding-api.open-meteo.com/v1/search?name={city}&count=1` → `r - Network accessline 30
https://api.open-meteo.com/v1/forecast?latitude={lat}&longitude={lon} - Network accessline 46
https://archive-api.open-meteo.com/v1/archive?latitude={lat}&longitude={lon} - Network accessline 55
https://air-quality-api.open-meteo.com/v1/air-quality?latitude={lat}&longitude={lon}¤t=pm2_5,us_aqi,european_aqi - Network accessline 67
https://api.weather.gov/alerts/active?area={two-letter state} ← e.g. area=CA
How the signals fit together
Facts from a static read of the files at the pinned commit. They describe what the code does, not why.
Some files could not be read; the facts below still hold.
Signal confidence: 205 facts
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:28High confidence
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:29High confidence
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:30High confidence
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:41High confidence
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:42High confidence
- Network access in code of the SKILL.mdapi-client-generator/SKILL.md:43High confidence
- Network access in prose of the SKILL.mdcitation-tracker/SKILL.md:56Low confidence
- Network access in prose of the SKILL.mdcitation-tracker/SKILL.md:58Low confidence
- Network access in code of the SKILL.mddocker-compose-wizard/SKILL.md:70High confidence
- Network access in code of the SKILL.mddocker-compose-wizard/SKILL.md:71High confidence
- Network access in code of the SKILL.mddocker-compose-wizard/SKILL.md:72High confidence
- Access to secrets in code of the SKILL.mdenv-doctor/SKILL.md:53High confidence
- Access to secrets in code of the SKILL.mdenv-doctor/SKILL.md:54High confidence
- Network access in code of the SKILL.mdenv-doctor/SKILL.md:112High confidence
- Network access in code of the SKILL.mdenv-doctor/SKILL.md:114High confidence
- Dangerous shell operation in prose of the SKILL.mdenv-doctor/SKILL.md:132Low confidence
- Dangerous shell operation in prose of the SKILL.mdenv-doctor/SKILL.md:134Low confidence
- Network access in code of the SKILL.mderror-handling/SKILL.md:84High confidence
- Network access in code of the SKILL.mderror-handling/SKILL.md:85High confidence
- Network access in code of the SKILL.mderror-handling/SKILL.md:86High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:31High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:43High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:44High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:45High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:53High confidence
- Network access in code of the SKILL.mdfinance-fx/SKILL.md:65High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:33High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:34High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:35High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:36High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:37High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:53High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:54High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:55High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:56High confidence
- Network access in code of the SKILL.mdfun-content/SKILL.md:57High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:32High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:46High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:54High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:67High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:80High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:81High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:85High confidence
- Network access in code of the SKILL.mdgeo-places/SKILL.md:87High confidence
- Network access in prose of the SKILL.mdgovernment-open-data/SKILL.md:19Low confidence
- Network access in code of the SKILL.mdgovernment-open-data/SKILL.md:32High confidence
- Network access in code of the SKILL.mdgovernment-open-data/SKILL.md:33High confidence
- Network access in code of the SKILL.mdgovernment-open-data/SKILL.md:34High confidence
- Network access in code of the SKILL.mdgovernment-open-data/SKILL.md:49High confidence
- Network access in code of the SKILL.mdgovernment-open-data/SKILL.md:57High confidence
and 155 more
Works with
| Agent | Status | Evidence | Install to |
|---|---|---|---|
| Claude Code | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .claude/skills/<skill-name>/ |
| Codex | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .agents/skills/<skill-name>/ |
| Cursor | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .cursor/skills/<skill-name>/ |
| Gemini CLI | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .gemini/skills/<skill-name>/ |
| GitHub Copilot | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .github/skills/<skill-name>/ |
| OpenCode | Compatible | Standard SKILL.md skills/1on1-agenda/SKILL.md | .opencode/skills/<skill-name>/ |
- Native The repository ships a manifest, folder or frontmatter field for this agent.
- Compatible A standard SKILL.md that this agent can read.
- Unknown No evidence either way.
Derived automatically from paths, manifests and frontmatter. Not tested on each agent.
