Individual skill
reflect-3_attempt_02
Not available
experiments/codex_gpt5_5_xhigh/reports/skills/task_group_011/reflect-3/reflect-3_attempt_02/SKILL.md
Source commit 5082a777379156a048881e94940d640702e5e09a
What was checked
- Matches the observed repository HEAD
- Yes
- Scanner rules are current
- No
- Repository scan complete
- No
- Skill scan complete
- No
- 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.
Standard SKILL.md format has not been attested at this commit. No installation command is available; inspect the source.
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.
- agent Repository topic
- ai Repository topic
- benchmark Repository topic
- llm Repository topic
- self-evolving Repository topic
- self-learning Repository topic
- skills Repository topic
Declared tags are partial or unavailable.
Files and excerpt
File metadata and permitted, redacted excerpts come from the recorded scan. Source links are pinned to that commit.
Read excerpt
# Credit Office Committee JSON Skill
Use this skill for credit-office evaluation tasks that ask for a committee-ready JSON answer from the public credit office API. The highest-risk failure mode is a structurally plausible answer that does not match the requested template. Treat the template, enum strings, and identifier spelling as controlling whenever they are present.
## Core Workflow
1. Read the prompt and any provided answer template before calculating. Preserve the template's top-level keys, nested object names, enum values, reason-code identifiers, owner identifiers, and required ordering.
2. Use only public API surfaces named in the prompt or manifest. Fetch the policy endpoint for calculation rules and the relevant benchmark endpoint before making recommendations.
3. Filter the population exactly as stated. Phrases such as "rated 3 or worse" and "rating 6 or worse" mean numeric ratings greater than or equal to that cutoff.
4. Keep calculations and presentation separate: first derive objective values, then map them into the template's allowed decision, posture, class, action, and reason-code fields.
5. Return one valid JSON object only. Do not add explanatory text outside JSON, and do not invent enum strings when the template gives a controlled vocabulary.
## Risk Rating Regrades
- Convert each available objective factor into its policy rating, then use the worst numeric rating among available DSCR, LTV/collateral, and delinquency factors.