OpenNeuro

You can access the OpenNeuro neuroimaging dataset via a dedicated MCP server that exposes GraphQL queries against the OpenNeuro API. This lets you explore datasets, snapshots, and file listings without needing separate API keys, using a consistent MCP client experience.

How to use

Connect your MCP client to the OpenNeuro MCP Server to run GraphQL queries, discover available fields through schema introspection, and fetch dataset information or file listings. You can use the remote MCP URL to connect directly, or run the server locally during development. Typical usage patterns include querying a dataset by ID to retrieve its metadata, listing files within a snapshot, and performing schema introspection to discover all supported query shapes.

How to install

Prerequisites you need before starting: Node.js and npm installed on your machine.

Clone the project, install dependencies, and run the development server.

git clone https://github.com/quentincody/open-neuro-mcp-server.git
cd open-neuro-mcp-server
npm install
npm run dev

Deployment and quickstart connections

You have two ways to connect to the OpenNeuro MCP Server from clients like Claude Desktop. Use the remote MCP URL for a cloud deployment, or run the server locally and point your client at the local address.

// Remote deployment via Cloudflare Workers
// Claude Desktop configuration (MCP server entry)
{
  "mcpServers": {
    "openneuro": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://open-neuro-mcp-server.quentincody.workers.dev/sse"
      ]
    }
  }
}

For local development, connect to the local server end point.

{
  "mcpServers": {
    "openneuro": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:8787/sse"
      ]
    }
  }
}

Example queries you can run

Use the available GraphQL tool to fetch data from OpenNeuro. Here is a typical query pattern you will execute against the MCP server to retrieve dataset information and file listings.

# Get dataset information
{\n  dataset(id: "ds000224") {\n    id\n    name\n    description\n    created\n  }\n}\n

Schema and capabilities

You can introspect the schema to discover all available queries and fields, then tailor your requests to fetch exactly the data you need.

{\n  __schema {\n    queryType {\n      name\n      fields {\n        name\n        description\n      }\n    }\n  }\n}\n

Where to run the server in production

If you deploy to Cloudflare Workers, the MCP server will be available at the deployed URL and can be connected by your clients using the same MCP configuration shape used for local and remote development.

Notes on access and licensing

OpenNeuro data is accessible publicly and does not require API keys for public datasets. The project is available under the MIT License with an academic citation requirement for publications that use this server. Ensure you attribute appropriately in academic work.

Usage summary

• Connect to the OpenNeuro MCP Server via remote URL or run it locally for development. - Use the GraphQL tool to query datasets, snapshots, and files. - Perform schema introspection to learn available fields. - Deploy to a cloud endpoint if you want to expose the MCP server to your clients.

Available tools

openneuro_graphql_query

Tool to execute GraphQL queries against the OpenNeuro API through the MCP server, enabling dataset lookups, file listings, and metadata retrieval.