Verified record · from the answer index
evaluate-run
skilltested · static-verifiedscore 50/100signed passport ✓published 52d ago · verified 2026-10-240 uses
Grade a completed factory run (a worktree + its chat) against your standards, cold and adversarially, so the system learns. Squashes the session transcript into a compact digest (prompts + corrections, the ordered tool timeline, the diff),
Declared tools (1)
evaluate-run
skill:uiverify/uiverify/evaluate-run · published by uiverify · source repository