review-criteria
Fifth stage of a guided AI security review. Walks an organisation through writing its own acceptance criteria into criteria.json. For each measured dimension it explains what the number means, how wide its interval is at the depth run, and what each kind of error costs in this deployment, and offers the organisation's own earlier results as reference points. Records who set each criterion, when, and whether results for the subject had been seen. Use before results are read where possible, and before review-report. Never proposes a value: the shipped template is empty and the tool has no defaults.
- License
- MIT
- Compatibility
- Requires python>=3.10 (standard library only). No network, no API keys.
Pinned to revision 3fb15e11d58f, so it is the text this page describes rather than whatever the author pushed since.
Files
- skills/review-criteria/SKILL.md
- skills/review-criteria/criteria.template.json
- skills/review-criteria/examples/criteria.example.json
- skills/review-criteria/lib/evalkit.py
- skills/review-criteria/scripts/criteria.py
- skills/review-criteria/scripts/expread.py
- skills/review-criteria/tests/test_criteria.py
Every link opens the file at its source, pinned to the revision this page describes.