Incidents¶

GET /v1/incidents¶

List incidents for your account. Returns up to 50 incidents, defaulting to the last 7 days.

curl "https://api.noctuary.io/v1/incidents?since=2026-05-01T00:00:00Z" \
  -H "Authorization: Bearer noc_ro_your_key"

Query parameters¶

Parameter	Format	Default	Description
`since`	ISO 8601	7 days ago	Only return incidents triggered after this time
`service`	string	—	Filter by service name

Response: 200 OK

[
  {
    "id": "f3a91b2c-...",
    "tenant_id": "5f0bfa6c-...",
    "service_id": "a8d4e91c-...",
    "triggered_at": "2026-05-31T02:14:41Z",
    "hypothesis": "Deploy sha=a4f2c1 to payments-api rolled out 4 minutes before onset",
    "recommended_action": "Roll back to sha=f8b3e22",
    "confidence": 0.94,
    "external_ref": "PD-12345",
    "external_source": "pagerduty",
    "posted_back": true,
    "resolved_at": null,
    "engineer_confirmed": null,
    "created_at": "2026-05-31T02:14:52Z"
  }
]

GET /v1/incidents/{id}¶

Get a single incident with full detail including the context snapshot.

curl https://api.noctuary.io/v1/incidents/{id} \
  -H "Authorization: Bearer noc_ro_your_key"

Response: 200 OK — same shape as list item, with context_snapshot populated.

The context_snapshot field contains the full state window at the time the incident was diagnosed — the exact evidence the LLM used to produce the hypothesis.

PUT /v1/incidents/{id}/feedback¶

Record whether an engineer confirmed or rejected the diagnosis. This feeds back into the causal vector store to improve future correlations.

curl -X PUT https://api.noctuary.io/v1/incidents/{id}/feedback \
  -H "Authorization: Bearer noc_ro_your_key" \
  -H "Content-Type: application/json" \
  -d '{"confirmed": true}'

Response: 204 No Content

Request body¶

Field	Type	Description
`confirmed`	boolean	`true` if the diagnosis was correct, `false` if wrong

Tip

Marking incidents as confirmed or wrong is the most valuable feedback you can give. Confirmed diagnoses are stored as causal patterns and improve accuracy for similar incidents in future.