GET /v1/health

Liveness and dependency status. Useful for monitoring and to confirm the service is reachable. This endpoint is not authenticated.

The API probes its upstream capabilities on a periodic schedule. When everything is healthy the response is just { "ok": true }. When a probe has failed, ok is false and a warnings array names the affected capabilities in plain language. The response never includes vendor names, hosts, or error detail.

The endpoint always answers 200 OK, even when a dependency is failing. That is deliberate: a dependency outage is not a reason to take a healthy API replica out of service. Inspect ok in the body for dependency health.

Request

GET /v1/health

Response 200 OK — healthy

{
  "ok": true
}

Response 200 OK — degraded

{
  "ok": false,
  "warnings": ["ai service is unhealthy"]
}

Field	Type	Description
ok	boolean	false once any dependency probe has failed, true otherwise.
warnings	string[]	Present only when a dependency probe has failed. One message per affected capability, e.g. "places service is unhealthy".

A capability is unhealthy when its most recent probe failed: a connection error, a timeout, a non-success response, or a response that did not look plausible. Treat this as "some capabilities may be unavailable" rather than a hard outage — the API may still answer with partial or fallback results.

The response is a sample from the replica that answers the request; different replicas may report different warnings while a dependency is unstable.

Example

curl -s https://api.overlookhospitality.com/v1/health