-
Notifications
You must be signed in to change notification settings - Fork 1
Pipeline Health Trend Analyzer with Proactive Auto-Intervention #649
Open
Description
Strategic Improvement
Pipeline vitals shows current health score but we don't track trends or intervene proactively. Build a trend analyzer that detects degrading health (2+ consecutive drops) and triggers diagnostic patrols BEFORE the next build fails. Current alerting (#634) is reactive; this makes the system proactive and self-healing.
Acceptance Criteria
- Store per-repo health scores in SQLite with timestamps (extend
sw-db.shschema) - Trend detection: flag repos with 2+ consecutive health drops or sudden >20% degradation
- Auto-intervention triggers: degrading health → run diagnostic patrol → inject findings into next build loop context
- Dashboard widget showing 7-day health trend sparkline per repo with up/down/stable indicators
shipwright health trendscommand showing repos sorted by trend direction (most degraded first)- Integration with daemon: before spawning build loop, check trend and boost effort level if degrading
Context
- Priority: P0
- Complexity: standard
- Generated by: Strategic Intelligence Agent
- Strategy alignment: P0: Reliability & Success Rate + P4: Observability