What problem does it solve?
This Skill continuously detects when connected integrations degrade or fail, and when scheduled routines stop running, run overdue, or fail to complete—so you can fix issues before they impact your work.
Core Features & Use Cases
- Integration health monitoring: Runs multi-layer checks per connected integration (credentials, tokens, scopes/capabilities, and functional reads) and flags broken vs degraded behavior.
- Routine health monitoring: Detects overdue reminders, missing execution records, stale or failed completions, and dependency-chain breakages between routines.
- Slack alerting with silence + reminders + restore notices: Sends first alerts on change, then repeats after a configurable silence window, and notifies again when systems are back to normal.
- Daily intent scan (optional): Conservatively scans recent conversations for routine-intent signals and can create missing Tier-1 routines or ask for missing details for Tier-2.
- Self-healing monitoring setup: Ensures the “Integration health check” routine exists (recreates if missing, deduplicates if duplicates exist).
Quick Start
Run the integration-health-check routine to write health snapshots to /agent/brain/integration-health/health-state.json and /agent/brain/integration-health/routine-health-state.json and to alert the admin Slack channel if anything is degraded or broken.