Grafana

You can run a dedicated MCP server that lets AI assistants interact with Grafana dashboards, datasources, alerts, incidents, and more. This Grafana MCP server is implemented in TypeScript/JavaScript and exposes a set of tools that your AI can call to manage Grafana resources safely through the MCP (Model Context Protocol).

How to use

Install and start the MCP server, then connect your MCP client. You will provide Grafana connection details via environment variables, and the MCP server will expose 43 tools for dashboard, datasource, Prometheus, Loki, incidents, alerts, on-call, and other Grafana operations. Use an MCP client of your choice (Claude Desktop, Claude Code, Cursor, Zed, Codex, etc.) and configure it to launch the MCP server as a local stdio process. Once connected, you can ask natural language questions like, “Show me dashboards matching CPU usage,” or “List all data sources and their types,” and the MCP server will execute the corresponding Grafana actions and return structured results.

How to install

Prerequisites: Node.js and npm must be installed on your machine. You will run the MCP server locally or install it globally for one-off usage.

Option 1: Install globally and run the MCP Grafana server.

npm install -g @leval/mcp-grafana
npx @leval/mcp-grafana

Environment setup for Grafana access

Create a configuration file or set environment variables to point the MCP server at your Grafana instance and authorize access.

GRAFANA_URL=https://your-grafana-instance.com
GRAFANA_SERVICE_ACCOUNT_TOKEN=glsa_xxxxxxxxxxxx
# Optional for debugging
DEBUG=true

Local development workflow

If you are developing locally, clone the project, install dependencies, and build before running.

git clone https://github.com/levalhq/mcp-grafana.git
cd mcp-grafana
npm install
npm run build

Available tools

search_dashboards

Search for dashboards by name or content across Grafana.

get_dashboard_by_uid

Retrieve the complete JSON of a dashboard by its UID.

get_dashboard_summary

Fetch metadata about a dashboard, such as title and last updated.

get_dashboard_property

Extract specific properties from a dashboard, like panel titles.

update_dashboard

Create or update a Grafana dashboard programmatically.

list_datasources

List all configured Grafana datasources.

get_datasource_by_uid

Get details for a datasource by its UID.

get_datasource_by_name

Retrieve a datasource configuration by name.

query_prometheus

Execute PromQL queries against configured Prometheus data.

list_prometheus_metric_names

List available Prometheus metric names.

list_prometheus_label_names

List label names for Prometheus metrics.

list_prometheus_label_values

Get values for a specific Prometheus label.

list_prometheus_metric_metadata

Describe metadata for a Prometheus metric.

query_loki_logs

Run LogQL queries against Loki logs.

query_loki_stats

Get statistics about Loki log streams.

list_loki_label_names

List label names available in Loki logs.

list_loki_label_values

Show values for a Loki log label.

find_error_pattern_logs

Analyze logs to identify recurring error patterns.

list_incidents

List active Grafana incidents.

get_incident

Get details for a specific incident.

create_incident

Create a new incident with given details.

add_activity_to_incident

Add notes or updates to an existing incident.