assistant-mcp

Connect AI coding agents to Grafana Cloud via an MCP server.

213|18|Updated Mar 17, 2026
One-click install
npx skills add https://github.com/grafana/skills --skill assistant-mcp
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: assistant-mcp
Source: https://github.com/grafana/skills/tree/main/skills/grafana-cloud/assistant-mcp
Command: npx skills add https://github.com/grafana/skills --skill assistant-mcp

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enable AI coding agents to access and control Grafana Cloud resources directly via the Model Context Protocol (MCP), removing context switching and manual API calls and allowing agents to run queries, inspect dashboards, and manage alerts from their coding environment.

Core Features & Use Cases

  • Transports: support for stdio (agent-spawned subprocess) and SSE (long-lived server) to fit local, team, and editor-based workflows.
  • Agent integrations: configuration examples and steps for Claude Code, Cursor, and VS Code MCP clients.
  • Tooling exposed: run PromQL, LogQL, and TraceQL queries; search and retrieve dashboards; list and inspect alert rules and incidents; manage fleet collectors and pipelines when enabled.
  • Use Case: have an agent retrieve the p95 latency for a service, search dashboards tagged "kubernetes", or list unhealthy collectors without leaving your editor.

Quick Start

Add a Grafana MCP server entry to your agent's MCP configuration with GRAFANA_URL and GRAFANA_API_KEY and run the mcp-grafana binary using stdio for single sessions or SSE for a shared server.

Frequently Asked Questions about assistant-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I connect an AI coding agent to Grafana Cloud?

To connect AI coding agents to Grafana Cloud, you configure an MCP server entry with GRAFANA_URL and GRAFANA_API_KEY, then run the mcp-grafana binary over stdio or SSE transports. This enables tool-based queries for metrics, logs, and traces.

Can I use Claude Code or Cursor to run PromQL and LogQL queries?

Yes, Claude Code and Cursor can run PromQL, LogQL, and TraceQL queries. By integrating with the Grafana MCP server, these agents execute queries and inspect dashboards directly within your coding environment without manual API calls.

What is needed to set up a Grafana MCP server for AI agents?

Setting up a Grafana MCP server requires a valid GRAFANA_URL, a Grafana Cloud service account token or API key, and the mcp-grafana binary. You add these credentials to your agent's MCP configuration to enable observability tooling.

Does Grafana MCP support searching dashboards and managing alert rules?

Yes, Grafana MCP supports searching and retrieving dashboards, listing alert rules, and inspecting incidents. It exposes these as tools for MCP-aware agents, allowing you to manage alerting and observability resources seamlessly.

What is the difference between stdio and SSE transports for MCP servers?

Stdio transport runs the mcp-grafana binary as an agent-spawned subprocess for single sessions, while SSE transport runs a long-lived shared server. Choose stdio for local workflows and SSE for team or editor-based environments.