Self-hosted docs
Observability
Observability starts with usable evidence, not dashboard volume.
Trace-first timelines
Each incident timeline keeps runtime rows, policy verdicts, and provider context together so teams can decide quickly.
- Example: Timeline shows runtime event with latency, cost, model metadata
- Example: Policy verdicts displayed inline with confidence scores and thresholds
- Example: Provider context includes status, pricing, release notes for affected services
- Request: GET /v1/incidents/inc-123/timeline
- Response: {incident_id: "inc-123", events: [{type: "runtime", timestamp: "...", latency_ms: 1800}, {type: "policy_verdict", verdict: "warn"}]}
- API Reference: See [Incident Timeline](/docs/api#incident-timeline) for full contract details
Operational exposure mapping
Cyiro maps upstream changes to active routes, showing where drift is already touching production.
- Example: Pricing change mapped to routes using affected model versions
- Example: Status degradation shown with blast radius of dependent routes
- Example: Release notes correlated with routes using changed features
- Request: GET /v1/drift/analysis?provider=anthropic&change_type=pricing
- Response: {affected_routes: ["prod-chat", "prod-summary"], blast_radius: 2, severity: "high"}
- API Reference: See [Drift Analysis](/docs/api#drift-analysis) for full contract details