Skip to content

Health and uptime monitoring

Prompy exposes a public GET /api/health endpoint for uptime monitoring. It checks database liveness, required configuration, and whether Pip is configured without returning secret values or calling Gemini.

The repository’s monitoring/ directory contains a pinned Gatus container, a Caddy HTTPS reverse proxy, and the endpoint configuration.

  1. Use an always-on Docker host; Gatus cannot run as a Vercel function.
  2. Point status.<your-domain> at the host.
  3. Copy monitoring/.env.example to .env and set the health token and alert credentials.
  4. Run docker compose up -d.

Gatus stores uptime history in the gatus-data volume. Keep this volume when updating the containers.

Unauthenticated requests return the overall status and boolean check results. Requests with Authorization: Bearer <HEALTH_CHECK_TOKEN> also receive latency and configuration booleans. The endpoint returns 200 when healthy and 503 when the database is unavailable or required configuration is missing.

Set NEXT_PUBLIC_STATUS_URL in the web deployment to link the public status page from the app footer and admin dashboard.