Menu

Individual skill

task-group-001-fewshot-attempt-01

0upvotes

Produce HarborCRM JSON answers for evaluation tasks that provide prompts, answer_template.json files, and HarborCRM API data. Use when reconciling event sponsorships, badge leads, campaign members, finance invoices, trade-show prospecting, or import batch hygiene into exact JSON output schemas with controlled enums, counts, dates, and CRM actions.

experiments/codex_gpt5_5_xhigh/reports/skills/task_group_001/demonstration_skill/demonstration_skill_attempt_01/demonstration-skill-attempt-01/SKILL.md

Read pinned source

Source commit 5082a777379156a048881e94940d640702e5e09a

What was checked

Matches the observed repository HEAD
Yes
Scanner rules are current
No
Repository scan complete
No
Skill scan complete
Yes
Observed behavior
Not checked yet
Last checked
Nov 6, 2026

These are observations at the recorded commit, not a guarantee of safety. Review the source before running it.

Install this skill

Choose an agent in your Stack or copy a command below. Commands fetch the pinned commit and copy only this skill directory. Existing destinations are refused. Copying a command does not install anything.

An install command is unavailable: current rules, complete repository and skill scans, an eligible behavior level and a permitted license are required.

Repository context

Repository stars (not skill ratings)
121
Repository last updated
Oct 29, 2026
License
Apache-2.0

Observed agent integration: No specific integration evidence. Standard SKILL.md installation does not imply verified compatibility.

Files and excerpt

File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.

Browse files
Read excerpt

# HarborCRM JSON Answer Procedures

## Operating Workflow

1. Read the user prompt and `input/payloads/answer_template.json` first. Treat the template as the contract: return exactly the requested top-level keys, field names, enum spellings, nullability, and ordering.
2. Use the API base URL supplied by the runner. Optionally check `/health`, then fetch only the endpoints needed for the task family.
3. Build normalized indexes before deciding actions:
   - Email: trim, lowercase, use empty string for blanks.
   - Phone: keep digits only; do not invent a country code.
   - Company/account: match exact names first, then normalized names, then email domain to CRM account `domain` when available.
   - Contacts: match normalized email first; if no email, use account plus normalized contact name or phone.
4. Derive decisions from API data, not from prose alone. Use the prompt only to choose the task family, target ID, required ranking, and any special sizing rules.
5. Emit JSON only. Validate mentally or with a JSON parser when possible. Do not add explanatory text, comments, or extra fields.
Back to top