{"slug":"agent-benchmark","source_name":"agent-benchmark","name":"Agent Benchmark","description":"Framework for measuring and tracking agent response quality over time. Detects regressions before they reach production. Use when evaluating agent changes, auditing quality, or establishing performance baselines.","version":1,"lift":{"pass_rate_delta_pts":63.64,"pass_rate_pct":77.3,"total_cases":22,"passed_cases":17,"tokens_delta_pct":121,"turns_delta_pct":0,"verdict":"mixed","benchmark_model":"gemini-3.6-flash","grading_method":"judged","completed_at":"2026-07-31T13:25:00.683540+00:00"},"skill_score":0.7727,"benchmark_models":[{"model":"gemini-3.6-flash","headline":true,"delta_pts":63.64,"with_pass_pct":77.3,"without_pass_pct":13.6,"tokens_delta_pct":121,"turns_delta_pct":0,"total_cases":22,"cases_aggregated":21,"verdict":"mixed","never_hurt":true,"completed_at":"2026-07-31T13:25:00.683540+00:00","run_id":"5a012e30-43c6-47dc-966d-53b009dc657a","latest_attempt":{"id":"5a012e30-43c6-47dc-966d-53b009dc657a","runner_model":"gemini-3.6-flash","overall_verdict":"mixed","completed_at":"2026-07-31T13:25:00.683540+00:00","gate":null},"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":0,"total_uses":0,"total_users":0,"manifest_hash":"928f6f7b098a371685a8c12738b506722268bd1421bf78b94cb7bfd523b590bc","install_name":"agent-benchmark","version_id":"c14929c0-b1a5-4bc2-9756-f92b1f91c5b5","body_sha256":"928f6f7b098a371685a8c12738b506722268bd1421bf78b94cb7bfd523b590bc","render_hash":"39293eaad94e695c215bcddc910993ba6f7f867b30f6dfd15bd8aee1e03d447b","raw_url":"https://app.decimal.ai/s/agent-benchmark/SKILL.md","scorecard_url":"https://app.decimal.ai/skills/agent-benchmark"}