Incident Analysis
Use when
- User asks why an incident happened or wants a causal chain.
- Response to operational regressions in latency, deadlocks, or replication lag.
Workflow
- Pull recent lock, query, and replication signal snapshots.
- Correlate temporal overlap and severity transitions.
- Build top incident hypotheses with confidence and remediation order.
- Return expected validation probes and rollback readiness.
Governance
- Do not trigger any destructive action.
- Mark confidence clearly before making any operational recommendation.
Output
timeline,incidentHypotheses,confidence,recommendedActions