Arthur MCPArthur MCP
Dashboard
How it worksFeaturesDocsEarly accessGitHubDashboard

How to Import the Grafana Dashboard

Get a ready-made view of Arthur's metrics — request rate, errors, latency, Node.js runtime metrics, and MCP tool/resource metrics — instead of building a dashboard from scratch.

Prerequisites: a running Prometheus instance scraping Arthur's /metrics endpoint, and a Grafana instance with that Prometheus added as a datasource. The repository's observability/ folder has local Docker Compose helpers for Prometheus, Grafana, and Tempo if you don't have your own stack yet.

Steps

  1. In Grafana, go to Dashboards → Import.

    Open Arthur's infrastructure observability page before switching to the external Grafana procedure

  2. Upload observability/grafana-dashboard.json from the Arthur MCP repository.

    Frame the Prometheus metrics that the Grafana dashboard consumes

  3. Select your Prometheus datasource when prompted.

  4. Import.

    Show the active exporter configuration before the tutorial continues in Grafana's external UI

Confirm it worked

The dashboard loads with panels for request rate, errors, latency, Node.js runtime metrics, MCP tool/resource metrics, and external HTTP metrics, all populated from your instance.

Good to know

  • If your API doesn't run on the default port, update observability/prometheus.yml so Prometheus scrapes the right address.

Related

Tutorial video