Dashboard context control
Search dashboards, retrieve summaries, versions, properties, panel queries, or targeted patches without always loading full JSON.
Documentation ↗MCP server / observability / source-linked
Grafana MCP
Grafana MCP is an open-source server that can run over stdio, SSE, or Streamable HTTP. It exposes dashboards, datasources, queries, alerts, incidents, annotations, Kubernetes resources, and documentation. The current server includes context-window-aware summary/property tools and explicit network security controls for HTTP deployment.
Before you choose.
Identify the implementation and the job it supports.
Grafana MCP is an open-source server that can run over stdio, SSE, or Streamable HTTP. It exposes dashboards, datasources, queries, alerts, incidents, annotations, Kubernetes resources, and documentation. The current server includes context-window-aware summary/property tools and explicit network security controls for HTTP deployment.
Setup and connection.
The documented interface, authentication and installation path.
- Connection / transport
- stdio by default; SSE or Streamable HTTP optional
- Authentication
- Grafana service-account token for the upstream Grafana instance; networked mode can require a separate bearer token for MCP callers.
Install with uvx, Docker, a binary, or Go; set GRAFANA_URL and GRAFANA_SERVICE_ACCOUNT_TOKEN. For network mode select -t streamable-http and set MCP_GRAFANA_SERVER_TOKEN.
Setup reference ↗Documented setup example. Review the source and credentials before running; this command was not executed for this article.
{"mcpServers":{"grafana":{"command":"uvx","args":["mcp-grafana"],"env":{"GRAFANA_URL":"http://localhost:3000","GRAFANA_SERVICE_ACCOUNT_TOKEN":"TOKEN"}}}}What it actually exposes.
Concrete capabilities with maintainer sources.
Datasource and query access
List/fetch datasources and optionally run panel queries; query examples can be enabled separately.
Documentation ↗Operational security
Validate Host/Origin headers, require caller bearer auth, expose health/metrics, and configure TLS for Streamable HTTP.
Documentation ↗Evidence and method.
What was checked and what was not tested.
HOW
Source method
Public maintainer documentation and repositories. This is not a hands-on product review.
WHEN
Last checked 2026-10-01
Setup examples are documented examples. Suggested trials and comparison advice are editorial guidance.
Sources support the adjacent factual statements. A source review does not establish uptime, performance, security or current account eligibility. Read the editorial policy.
Try a specific task.
A bounded evaluation you can reproduce in your own environment.
Dashboard-to-incident loop
Suggested evaluation, not an observed test result.
- Connect read-only credentials and select a bounded dashboard or datasource scope.
- Ask for a compact dashboard summary, anomalous panels, and linked logs or incidents.
- Review any proposed dashboard patch or query before enabling write/query tools.
What to inspect: A focused incident context packet that avoids flooding the model with full dashboard JSON. Keep the source links beside each claim when publishing the record. Proposed workflow; not tested by clawbot.ai.
Workflow reference ↗Permissions and limitations.
Read these constraints before granting access.
Networked modes require explicit caller authentication and host/origin controls; without them the server is unsafe to expose beyond a trusted boundary.
Boundary reference ↗Some tools such as panel queries and examples are disabled by default, and Grafana 9+ is required for full functionality.
Boundary reference ↗Decide by workflow.
Make the expected output explicit before you connect.
A focused incident context packet that avoids flooding the model with full dashboard JSON. Keep the source links beside each claim when publishing the record. Proposed workflow; not tested by clawbot.ai.
Keep the task inputs, tool calls, output and any errors together when you evaluate. The trial above is a proposal, not a recorded result.
Compare the alternatives.
Specific reasons to choose another route, with the tradeoff shown.
| Alternative | Choose it when | Tradeoff |
|---|---|---|
| Sentry MCP | Choose for issue, trace, and release-centric debugging with hosted OAuth. | Grafana offers broader dashboards and datasource control but needs instance credentials. Documentation ↗ |
| Vercel MCP | Choose for Vercel deployment logs and runtime errors. | Vercel is narrower; Grafana spans multiple observability backends. Documentation ↗ |
Questions before setup.
Answers specific to this implementation.
Can I expose it over HTTP?
Yes, use SSE or Streamable HTTP, but set caller auth and allowed hosts/origins before exposing the listener.
Documentation ↗
Why use summary tools?
Large dashboard JSON can consume context; summary/property tools retrieve bounded slices.
Documentation ↗
Source trail.
Maintainer documentation used in this record.