{"slug":"pramoddutta-ai-agent-evaluation","source_name":"pramoddutta/AI Agent Evaluation","name":"Pramoddutta/AI Agent Evaluation","description":"Comprehensive evaluation patterns for AI agents including multi-turn conversation testing, LLM-as-judge frameworks, benchmark suites, regression detection, and systematic eval pipelines for measuring agent quality and safety.","version":1,"lift":{"pass_rate_delta_pts":21.74,"pass_rate_pct":78.3,"total_cases":23,"passed_cases":18,"tokens_delta_pct":247.7,"turns_delta_pct":0,"verdict":"mixed","benchmark_model":"gemini-3.6-flash","grading_method":"judged","completed_at":"2026-09-18T23:16:29.937818+00:00"},"skill_score":0.7826,"benchmark_models":[{"model":"gemini-3.6-flash","headline":true,"delta_pts":21.74,"with_pass_pct":78.3,"without_pass_pct":56.5,"tokens_delta_pct":247.7,"turns_delta_pct":0,"total_cases":23,"cases_aggregated":23,"verdict":"mixed","never_hurt":false,"completed_at":"2026-09-18T23:16:29.937818+00:00","run_id":"b4c5bb15-fbac-487d-8e07-6e7d9850a60d","latest_attempt":{"id":"b4c5bb15-fbac-487d-8e07-6e7d9850a60d","runner_model":"gemini-3.6-flash","overall_verdict":"mixed","completed_at":"2026-09-18T23:16:29.937818+00:00","gate":null},"version_number":1,"is_latest_version":true,"gate":null}],"trust":{"skill_safety":"passed","safety_status":"clean","intent_verdict":"safe","content_status":"clean","indexable":true},"license":"MIT","install_count":0,"total_uses":13,"total_users":1,"manifest_hash":"9f66904b6f7b2a69a435573e55e4971b55d0b1c8782c529a9e60362fd41f5fef","install_name":"pramoddutta-ai-agent-evaluation","version_id":"1bbb9691-e9e4-4a1f-b3aa-dbb167699696","body_sha256":"9f66904b6f7b2a69a435573e55e4971b55d0b1c8782c529a9e60362fd41f5fef","render_hash":"5bdd7da87e8207c4b4c69feb3b2ed2efab69a078b187800a72539f302d0b1069","raw_url":"https://app.decimal.ai/s/pramoddutta-ai-agent-evaluation/SKILL.md","scorecard_url":"https://app.decimal.ai/skills/pramoddutta-ai-agent-evaluation"}