You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Three */5 scheduled tasks stopped dispatching with no run rows #256
backfill-skill-assets, drain-skill-dirty, and sync-reviewed-skill-repos stopped dispatching at 2026-09-18T20:20Z with zero scheduled_runs rows and nothing detects or re-invokes them, so their queues silently back up.
The daily-checkin routine opened this issue automatically on its 2026-09-18T21:40:00.000Z run. It is not Harlan's own report. Close it to reject the proposal, and the reason you give stops it being offered again.
Summary: Production incident confirmed by the daily-checkin: backfill-skill-assets, drain-skill-dirty, and sync-reviewed-skill-repos (all */5) stopped producing scheduled_runs rows at 2026-09-18T20:20Z, while sibling */5 task ai-ready:cron kept runn
Next action: Implement server/tasks/scheduled-cadence-watchdog.ts: query scheduled_runs per SCHEDULE_POLICY task, flag tasks whose latest run exceeds maxSilenceSeconds (or has no rows past activeFromSeconds), skip any with a live 'started' run inside maxRuntimeSeconds, and re-invoke the stalled tasks via Nitro runTask; add its observed policy entry to shared/schedule-policy.ts, run pnpm cron:docs to regenerate CRON.md, add a spec test following layers/registry/server/tasks/drain-skill-dirty.spec.ts, deploy, then run the issue's verify curl against /api/internal/checkin.
Activity
harlan-github-agent commentedon Sep 18, 2026
🤖 ISSUE TRIAGE