Observability

Arthur MCP exposes the operational signals a team needs to run it in production, so problems surface before someone has to go digging.

What you can do

• Check instance health at a glance — is the app up, ready, and live?
• Read structured logs with correlation IDs, so a single request can be traced end to end.
• Pull Prometheus metrics for dashboards and alerting.
• Enable optional OpenTelemetry tracing for deeper request-level visibility.
• Import a ready-to-use Grafana dashboard, backed by local Prometheus, Grafana, and Tempo helper configuration.

How to use it

1. Go to Observability to see the current, live status of health checks, metrics, and tracing on this instance.
2. Point your monitoring stack at the exposed health and metrics endpoints.
3. Import the provided Grafana dashboard if you want a ready-made view instead of building one from scratch.

Good to know

• Health and metrics endpoints are intentionally public, since monitoring infrastructure needs to reach them without authenticating.
• Related: Settings for the read-only Observability tab, Error Tracking and Audit Logs for two other kinds of visibility.