eval
Build and run evaluations for an AI agent or LLM application, including harnesses, failure-based tests, benchmarks and shared evaluators. Also analyze evaluation coverage or design scenarios without requiring implementation or execution. Repair broken tests or mock unreachable dependencies. Use improve for changing the agent or comparing changes; use coding-evals for agent-neutral coding tasks, including agents developed here.
Pinned to revision 374f8ed90083, so it is the text this page describes rather than whatever the author pushed since.
Files
- skills/eval/SKILL.md
- skills/eval/references/benchmarks.md
- skills/eval/references/harbor-notes.md
- skills/eval/references/hardening.md
- skills/eval/references/harness.md
- skills/eval/references/measure.md
- skills/eval/references/mocks.md
- skills/eval/references/review-harness.md
- skills/eval/references/review-test.md
- skills/eval/references/shared-evaluators.md
- skills/eval/references/test-design.md
Every link opens the file at its source, pinned to the revision this page describes.