# Grafana > Query dashboards, Prometheus/Loki datasources, alerts, and incidents from an AI agent — official. [Canonical HTML page](https://top-mcps.com/mcp/grafana) · [server.json](https://top-mcps.com/mcp/grafana.json) · [methodology](https://top-mcps.com/about/methodology) ## Install ### Claude Desktop — `claude_desktop_config.json` Paste under mcpServers. Fully quit and reopen Claude after editing. ```json { "mcpServers": { "grafana": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### Claude Code — `CLI or .mcp.json` Run from your repo. Commit .mcp.json to share with your team. ```shell # export GRAFANA_URL=https://myinstance.grafana.net # export GRAFANA_SERVICE_ACCOUNT_TOKEN=glsa_your_token claude mcp add grafana -- docker run --rm -i -e GRAFANA_URL -e GRAFANA_SERVICE_ACCOUNT_TOKEN grafana/mcp-grafana -t stdio ``` ### Cursor — `.cursor/mcp.json` Global path: ~/.cursor/mcp.json. Reload window after editing. ```json { "mcpServers": { "grafana": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### VS Code — `.vscode/mcp.json` VS Code uses the "servers" key (not "mcpServers"). ```jsonc { "servers": { "grafana": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### Windsurf — `~/.codeium/windsurf/mcp_config.json` Open via Cascade → hammer icon → Configure. ```json { "mcpServers": { "grafana": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### Cline — `cline_mcp_settings.json` Open via the Cline sidebar → MCP Servers → Edit. ```json { "mcpServers": { "grafana": { "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### Continue — `~/.continue/config.json` Continue uses modelContextProtocolServers with a transport block. ```json { "experimental": { "modelContextProtocolServers": [ { "transport": { "type": "stdio", "command": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ], "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } ] } } ``` ### Codex CLI — `~/.codex/config.toml` Codex uses TOML. Each server is a [mcp_servers.] subtable. ```shell # ~/.codex/config.toml [mcp_servers.grafana] command = "docker" args = [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio", ] env = { GRAFANA_URL = "${GRAFANA_URL}", GRAFANA_SERVICE_ACCOUNT_TOKEN = "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } ``` ### Zed — `~/.config/zed/settings.json` Zed calls them "context_servers". Settings live-reload on save. ```jsonc { "context_servers": { "grafana": { "command": { "path": "docker", "args": [ "run", "--rm", "-i", "-e", "GRAFANA_URL", "-e", "GRAFANA_SERVICE_ACCOUNT_TOKEN", "grafana/mcp-grafana", "-t", "stdio" ] }, "env": { "GRAFANA_URL": "${GRAFANA_URL}", "GRAFANA_SERVICE_ACCOUNT_TOKEN": "${GRAFANA_SERVICE_ACCOUNT_TOKEN}" } } } } ``` ### ChatGPT — `ChatGPT → Apps directory` Grafana doesn't ship a hosted HTTPS endpoint today. ChatGPT supports remote MCP servers only — to use this server in ChatGPT you'll need to deploy it to a public HTTPS URL first (e.g. via Cloudflare Workers or Vercel) or wait for an official remote build. ```none ``` ## At a glance - **Maintainer:** Grafana Labs - **Transport:** stdio, SSE, Streamable HTTP - **Auth model:** API key - **Required secrets:** GRAFANA_URL, GRAFANA_SERVICE_ACCOUNT_TOKEN - **Supported clients:** Claude, Cursor, VS Code, Any MCP-compatible client, Grafana 9.0+ - **License:** Apache-2.0 - **Language:** Go - **Latest version:** latest - **Last verified:** 2026-07-16 - **GitHub stars:** 3,246 (fetched 2026-07-17T12:42:37.493Z) - **Score:** 92/100 (rubric 2026-04 — see https://top-mcps.com/about/methodology) - **Source:** https://github.com/grafana/mcp-grafana ## Quick answer **What it does.** Exposes Grafana operations as MCP tools: dashboard search/edit, datasource queries (Prometheus, Loki, and 10+ others), alert-rule inspection, and Incident/OnCall/Sift workflows. **Best for:** - Incident investigation across metrics + logs - Querying Prometheus and Loki via one MCP - Dashboard search and editing - Alert-rule inspection - Grafana Incident / OnCall / Sift workflows **Not for:** - Grafana versions before 9.0 - Direct datasource access without Grafana - Instances where a service-account token cannot be issued ## Description Grafana's official MCP (Apache-2.0, ~3.2k stars) gives an agent read/write access to a Grafana instance: search and edit dashboards, query Prometheus and Loki datasources, inspect alert rules, and drive Incident, OnCall, and Sift. It turns "what fired at 2am" into a single grounded question. ## Why it matters Observability data is only useful if you can ask it questions. A Grafana MCP lets an agent pull the metric, correlate the log, and read the alert without a human pivoting across tabs — the core of AI-assisted incident response. ## Key features - Dashboards + 12 datasource types (Prometheus, Loki, ClickHouse, CloudWatch…) - Alerting, Incident, OnCall, Sift tools - stdio, SSE, and streamable-http transports - Service-account token auth - Official Grafana Labs, Apache-2.0, Go ## FAQ ### What auth does it need? A Grafana service-account token (`GRAFANA_SERVICE_ACCOUNT_TOKEN`) plus your instance `GRAFANA_URL`. The deprecated `GRAFANA_API_KEY` still works. ### Which datasources can it query? Prometheus and Loki are first-class, plus ClickHouse, CloudWatch, Elasticsearch/OpenSearch, InfluxDB, Snowflake, Athena, and more. ### Minimum Grafana version? Grafana 9.0 or later for full tool coverage. ## Changelog - **2026-07-16** — Refreshed install snippets and fact sheet; verified for 2026. - **2025-04-01** — Initial directory listing.