{"definition_raw":"---\nid: skill-invisibility-weekly\ntitle: Weekly invisible-skills detector\nschedule: \"0 9 * * 1\"\ntimeout: 900\nretry: false\nenabled: true\nnotify_on: failure\nrun_as: shell\ncommand: \"python3 /home/lucienne/workspace/scripts/skill_audit.py\"\ntags: [skills, audit, observability]\nruntime_profile: direct_python\n---\n**OVERRIDES runtime profile:** uses `direct_python` (plain Python, no model) because the command chain never invokes the `claude` CLI or any LLM API \u2014 pure-infra task; scheduler provider env injection is a no-op (MC-4942 U12 sweep).\n\n# Invisible-skills detector (MC-797)\n\nComplements `skill-audit-weekly` (frontmatter/credentials lint). This task\nanswers a different question: of the ~120 skills we have, which ones never\ntrigger? The failure mode caught here is Atlas/GBrain (commit 62b0192) \u2014\na skill existed, looked healthy, but silently never matched real user\nphrasing.\n\nRuns Monday 09:00 SAST. Writes `~/workspace/reports/skill-audit-YYYY-MM-DD.html`\n(surfaces on the MC dashboard under /reports). Telegrams Elmar only when a\npreviously-active skill has gone silent or a near-miss recurs 3+ times.\n","id":"skill-invisibility-weekly","last_run":{"duration_s":35.197345,"log_path":"/home/lucienne/workspace/logs/task-runs/skill-invisibility-weekly/393166.log","output":"Wrote /home/lucienne/workspace/reports/skill-audit-2026-06-08.html\nTelegram notification sent.\n","started_at":"2026-06-08T09:06:06.960610+02:00","status":"completed"},"next_run":"2026-06-15 09:00","next_run_iso":"2026-06-15T09:00:00+02:00","runs":[{"duration_s":35.197345,"finished_at":"2026-06-08T09:06:42.160968+02:00","id":393166,"log_path":"/home/lucienne/workspace/logs/task-runs/skill-invisibility-weekly/393166.log","output":"Wrote /home/lucienne/workspace/reports/skill-audit-2026-06-08.html\nTelegram notification sent.\n","started_at":"2026-06-08T09:06:06.960610+02:00","status":"completed","task_id":"skill-invisibility-weekly","task_name":"Weekly invisible-skills detector"}],"runs_limit":20,"schedule":"0 9 * * 1","schedule_label":{"description":"Mon at 09:00","is_custom":false,"label":"Weekly (Mon)","sort":6,"sort_time":"09:00"},"stats":{"avg_duration":35.197345,"completed":1,"failed":0,"timeout":0,"total":1},"task":{"_description":"**OVERRIDES runtime profile:** uses `direct_python` (plain Python, no model) because the command chain never invokes the `claude` CLI or any LLM API \u2014 pure-infra task; scheduler provider env injection is a no-op (MC-4942 U12 sweep).\n\n# Invisible-skills detector (MC-797)\n\nComplements `skill-audit-weekly` (frontmatter/credentials lint). This task\nanswers a different question: of the ~120 skills we have, which ones never\ntrigger? The failure mode caught here is Atlas/GBrain (commit 62b0192) \u2014\na skill existed, looked healthy, but silently never matched real user\nphrasing.\n\nRuns Monday 09:00 SAST. Writes `~/workspace/reports/skill-audit-YYYY-MM-DD.html`\n(surfaces on the MC dashboard under /reports). Telegrams Elmar only when a\npreviously-active skill has gone silent or a near-miss recurs 3+ times.","_file":"skill-invisibility-weekly.md","_path":"/home/lucienne/workspace/tasks/skill-invisibility-weekly.md","command":"python3 /home/lucienne/workspace/scripts/skill_audit.py","enabled":true,"id":"skill-invisibility-weekly","notify_on":"failure","retry":false,"run_as":"shell","runtime_profile":"direct_python","schedule":"0 9 * * 1","tags":["skills","audit","observability"],"timeout":900,"title":"Weekly invisible-skills detector"}}