TrueRAG

You set up and run a Model Context Protocol (MCP) server that provides access to GraphQL-backed policies. This server lets MCP clients query and interact with policy data through a GraphQL API, powered by the Python MCP SDK and the GQL library. It’s useful when your applications need structured, policy-driven context from a centralized source while remaining decoupled from direct GraphQL client logic.

How to use

You use an MCP client to connect to the GraphQL Policies API exposed by this server. Once the server is running, your client can issue policy queries and receive structured responses that integrate with your application's decision-making and context assembly. You can run the server locally, then point your MCP client at the local or remote endpoint to fetch, filter, and apply policy data as part of your prompts or automated workflows.

How to install

Prerequisites you need before installation are a working Python runtime, a UNIX-like shell (macOS/Linux) or PowerShell on Windows, and Git for cloning repositories.

# 1) Clone the MCP server repository
git clone https://github.com/Ad-Veritas/mcp-server-trueRAG.git
cd mcp-server-trueRAG

# 2) Ensure you have uv installed
uv --version

If uv is not installed, install it using these commands. Choose the command appropriate for your operating system.

# On macOS and Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# On Windows
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Create the required environment configuration for the GraphQL API you’ll connect to. Place these values in a .env file at the root of the repository.

GRAPHQL_API_KEY = "{your_api_key}"
GRAPHQL_ENDPOINT = "{your_graphql_endpoint}"

Additional configuration and run notes

To run the MCP server with MCP clients such as Claude Desktop, you will start the server using uv with the directory that contains the MCP server code. The following config demonstrates how Claude can launch the server.

"shipping-policies": {
      "command": "uv",
      "args": [
        "--directory",
        "{path_to_mcp_server}/mcp-server-trueRAG",
        "run",
        "fastmcp",
        "run",
        "server.py"
      ]
    }

Troubleshooting and notes

If you encounter connection issues, verify that GRAPHQL_API_KEY and GRAPHQL_ENDPOINT are correctly defined in your .env and that the GraphQL service is reachable at the endpoint you configured. Ensure uv is installed and accessible in your PATH, and that you are running the command from the repository root or a path that resolves the --directory argument correctly.

Security and environment considerations

Treat the GraphQL API key as a sensitive credential. Store it securely, restrict access to the environment file, and rotate keys periodically according to your security policy.