Grafana MCP Server

What the Grafana MCP server does

How models use it and what it is built for.

The Grafana MCP server connects through your Grafana instance rather than to each datasource separately, which means one credential and one endpoint covers everything Grafana already knows how to reach. On the metrics side an agent can list datasources, discover metric names and label values, and run instant or range PromQL queries. On the logs side it can query Loki with LogQL, list label names and values, and fetch log statistics. It can search dashboards by title, fetch a dashboard by UID to read its panels, and update or create dashboards. Incident response is covered too: list and create incidents, read alert rules and their current state, and run Sift investigations that look for error patterns or slow requests automatically. A generate_deeplink tool returns a real Grafana URL so a human can pick up where the agent left off. Tools are grouped into categories you can disable individually, which keeps the tool list manageable.

Tools the Grafana MCP server exposes

Typical tools an AI model can call. Exact names vary by version.

• query_prometheus — run instant or range PromQL queries against a datasource
• list_prometheus_metric_names / label values — discover what is actually being recorded
• query_loki_logs / query_loki_stats — LogQL queries and log volume statistics
• search_dashboards / get_dashboard_by_uid / update_dashboard — read and edit dashboards
• list_datasources / get_datasource_by_uid — enumerate what the instance can query
• list_alert_rules / get_alert_rule_by_uid — alert definitions and current state
• list_incidents / create_incident — Grafana Incident workflows
• find_error_pattern_logs / find_slow_requests — Sift investigations
• generate_deeplink — hand a human a real Grafana URL for the view being discussed

Connecting to Grafana

Taken from the official Grafana documentation — see Grafana MCP server — official documentation for the full reference.

{
  "mcpServers": {
    "grafana": {
      "command": "uvx",
      "args": ["mcp-grafana"],
      "env": {
        "GRAFANA_URL": "https://myorg.grafana.net",
        "GRAFANA_SERVICE_ACCOUNT_TOKEN": "glsa_YOUR_TOKEN"
      }
    }
  }
}

{
  "mcpServers": {
    "grafana": {
      "command": "docker",
      "args": [
        "run", "--rm", "-i",
        "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN",
        "mcp/grafana", "-t", "stdio"
      ],
      "env": {
        "GRAFANA_URL": "https://myorg.grafana.net",
        "GRAFANA_SERVICE_ACCOUNT_TOKEN": "glsa_YOUR_TOKEN"
      }
    }
  }
}

Example prompts to try

Copy any of these into MCP Agent Studio after connecting.

• What was p99 latency on the checkout service over the last six hours?

• Find the error pattern in the payments logs since the deploy at 14:00.

• Which alert rules are currently firing, and which dashboards cover those services?

• Show me request rate by pod for the last hour and flag anything anomalous.

Models on MCP Playground

This is not a single-model product: you get the same MCP connection with 60+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.

Check an AI agent can actually use the Grafana MCP server

Listing tools proves the server is reachable, not that a model can work with it. Evals go further: they read every tool on the server, write a test suite from its real schemas, and run it — code decides pass/fail on the responses (schema conformance, error codes, pagination, result caps) while a scoring model grades plain-English tasks driven through the tools.

Grafana MCP server — FAQ

Common questions about connecting, scoping and using it safely.