{"slug":"eval-harness","source_name":"eval-harness","name":"Eval Harness","description":"Formal evaluation framework for Claude Code sessions implementing eval-driven development (EDD) principles","version":1,"lift":{"pass_rate_delta_pts":54.55,"pass_rate_pct":54.5,"total_cases":22,"passed_cases":12,"tokens_delta_pct":43.7,"turns_delta_pct":0,"verdict":"mixed","benchmark_model":"gemini-3.6-flash","grading_method":"judged","completed_at":"2026-07-27T16:43:50.752950+00:00"},"skill_score":0.7727,"benchmark_models":[{"model":"gemini-3.6-flash","headline":true,"delta_pts":54.55,"with_pass_pct":54.5,"without_pass_pct":0,"tokens_delta_pct":43.7,"turns_delta_pct":0,"total_cases":22,"cases_aggregated":21,"verdict":"mixed","never_hurt":true,"completed_at":"2026-07-27T16:43:50.752950+00:00","run_id":"c77b649c-9d57-4f16-be43-611d2e2d23db","version_number":1,"is_latest_version":true,"gate":null}],"trust":{"skill_safety":"passed","safety_status":"clean","intent_verdict":null,"content_status":"clean","indexable":true},"license":"MIT","install_count":1,"manifest_hash":"f38da682e0f9dbcc9a0ebdff8b34b3b81276f6303b6b7d634fcda9a151943953","raw_url":"https://app.decimal.ai/s/eval-harness/SKILL.md","scorecard_url":"https://app.decimal.ai/skills/eval-harness"}