Multi MCP

What is Multi MCP?

An MCP server that proxies multiple MCP servers, acting as a single entry point for clients. It is designed for developers who want to aggregate, authenticate, or namespace multiple MCP tools and resources.

How to use Multi MCP?

Clone the repository, install dependencies with uv pip install -r requirements.txt, and start the server with hypercorn src.main:app. Configure it by setting the CONFIG environment variable to a JSON file following VSCode’s MCP configuration format (examples are provided). Access the server via /sse for SSE mode or /mcp for streamable HTTP mode.

Key features of Multi MCP

• Proxies multiple MCP servers behind one endpoint.
• Supports SSE and streamable HTTP transports.
• Configurable transport via TRANSPORT environment variable.
• Optional authentication with AUTH_TOKEN.
• Namespaces tools and resources with USE_NAMESPACE.
• Debug mode via DEBUG environment variable.

Use cases of Multi MCP

• Expose several standalone MCP servers as one unified server.
• Add authentication to an MCP tool collection.
• Organize tools from different providers under separate namespaces.
• Switch between stdio, SSE, or HTTP transports without changing clients.

FAQ from Multi MCP

What dependencies are required?

Python environment with uv (for virtual environment and package management), plus hypercorn to run the server.

Where is the configuration file located?

The path is set by the CONFIG environment variable; if not set, it defaults to ./examples/config/mcp_tools.json.

What transport protocols are supported?

The TRANSPORT variable can be set to stdio, sse, or http (streamable HTTP).

How is authentication handled?

Set the AUTH_TOKEN environment variable to enable token-based authentication. If not set, authentication is disabled.

What is the purpose of the USE_NAMESPACE flag?

When enabled, tools and resources from different config entries are prefixed with a namespace to avoid name collisions.