{"slug":"sonichi-voice-agent-test-harness","source_name":"sonichi/voice-agent-test-harness","name":"Sonichi/Voice Agent Test Harness","description":"Drive a fixed suite of spoken tests against a voice agent (\"subject\") from a co-located machine (\"prober\"), measure response latency / clarity / accuracy, diff against baseline, and report to the owner over Telegram.","version":1,"lift":{"pass_rate_delta_pts":60.87,"pass_rate_pct":82.6,"total_cases":23,"passed_cases":19,"tokens_delta_pct":-5.3,"turns_delta_pct":0,"verdict":"mixed","benchmark_model":"gemini-3.6-flash","grading_method":"judged","completed_at":"2026-08-23T21:11:31.523397+00:00"},"skill_score":0.8261,"benchmark_models":[{"model":"gemini-3.6-flash","headline":true,"delta_pts":60.87,"with_pass_pct":82.6,"without_pass_pct":21.7,"tokens_delta_pct":-5.3,"turns_delta_pct":0,"total_cases":23,"cases_aggregated":22,"verdict":"mixed","never_hurt":false,"completed_at":"2026-08-23T21:11:31.523397+00:00","run_id":"32ff43d0-1dd3-48c9-b1a9-601f03de4962","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,"manifest_hash":"a6fa722b6ee1bcdc8150adc23094bdbe01c883cf37c890499a066e42ce0eda1b","raw_url":"https://app.decimal.ai/s/sonichi-voice-agent-test-harness/SKILL.md","scorecard_url":"https://app.decimal.ai/skills/sonichi-voice-agent-test-harness"}