Installing the FusionReactor Cloud MCP

To connect your AI tooling to FusionReactor Cloud using MCP, follow the steps below.

Step 1: Generate a FusionReactor API Key

You’ll need an API key so your tooling can authenticate with the FusionReactor MCP.

Info

Step 2: Install and configure the MCP Adapter

FusionReactor Cloud MCP is based on the Grafana MCP adapter.

Here’s an example configuration for connecting Claude Desktop to FusionReactor Cloud using the MCP adapter.

The simplest way to get started is by using the prebuilt Grafana MCP Docker image, together with SSE (Server-Side Event) streams. Grafana provides additional documentation for alternative methods.

Prerequisites

Node/NPM must be installed, particularly the npx command.
- Test: Run npx mcp-remote in a shell. You should see a Usage: message.
Docker must be installed and functioning.
- Test: Run docker ps in a shell to confirm it's working.
Claude Desktop, Claude Code, or another MCP host must be installed.
- This example uses Claude Desktop.
A FusionReactor Cloud API Key is required.
- Generate an FR API key

Procedure

Start the Docker container:

docker run --rm -d -p 8000:8000 -e GRAFANA_URL=https://api.fusionreactor.io/g -e GRAFANA_API_KEY=<FR API KEY> mcp/grafana
- The docker will run and detach.
- If you need to kill it later, use docker ps to find its name or container ID, then use docker kill <id> to stop it.

Configure Claude Desktop:

Go to: Claude → Settings → Developer → Edit Config
Add the MCP definition (example provided in next section)

{
  "mcpServers": {
    "grafana": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:8000/sse"
      ]
    }
  }
}

Restart Claude Desktop.
You should then be able to use this MCP in conversation with the AI:

Additional notes

This MCP interface is read-only – it allows querying data, not modifying it.
See the features section of Grafana MCP for a full list of supported functionality.