Apollo MCP Server

Apollo MCP Server allows you to expose your existing GraphQL APIs as tools for AI assistants, enabling them to query real-time data. It integrates seamlessly with AI clients like Claude Desktop, transforming GraphQL operations into AI-accessible functions.

What it does

This server acts as a bridge, allowing large language models (LLMs) to understand and execute GraphQL queries. It takes your GraphQL schema and defined operations, and presents them to an MCP client as a set of callable tools. For example, you can expose an operation to fetch astronaut data, and an AI assistant can then use this "tool" to answer questions like "Who are the astronauts currently in space?".

How to use

1. Clone the repository: git clone https://github.com/apollographql/apollo-mcp-server.git cd apollo-mcp-server
2. Install Rover CLI: Ensure you have Apollo Rover CLI v0.35 or later.
3. Run the server: Use the rover dev command with your supergraph and MCP configuration files.

   rover dev --supergraph-config ./graphql/TheSpaceDevs/supergraph.yaml \
     --mcp ./graphql/TheSpaceDevs/config.yaml
   

   This command starts a local graph and an MCP Server, typically on http://127.0.0.1:5000/mcp using Streamable HTTP transport.
4. Connect an MCP Client: Configure your AI assistant (e.g., Claude Desktop) to connect to the running MCP Server. For Claude, you would add a configuration like this to its claude_desktop_config.json:

   {
     "mcpServers": {
       "thespacedevs": {
         "command": "npx",
         "args": [
           "mcp-remote",
           "http://127.0.0.1:5000/mcp"
         ]
       }
     }
   }
   

5. Test: Ask your AI assistant questions that leverage the exposed GraphQL tools, such as "Who are the astronauts currently in space?".

Implementation Details

Apollo MCP Server leverages GraphQL Federation to combine multiple subgraphs into a single supergraph. It then exposes pre-built GraphQL operations from this supergraph as MCP tools. It supports Streamable HTTP and STDIO transports.