✓ All systems operational

Checked every five minutes against the database every request depends on. Last response 661ms.

How this is measured, and what it cannot show. A check runs every five minutes inside the deployment and exercises the database every request depends on — a process that is running but cannot reach Postgres is recorded as failing, not as healthy. Because the recorder runs inside the deployment, a total outage stops the recording rather than logging a failure: it appears as a gap, and this page reports an unconfirmed state instead of green. Anyone wanting independent verification can poll /api/health, which returns 503 when the service cannot serve.