MCP Composer

What is MCP Composer?

MCP Composer is a gateway service that centrally manages all your MCP servers. It consolidates multiple MCP servers and opens independent ports with different combinations of servers and tools for each service (such as AI agents or tools) that requires access to MCP servers.

How to use MCP Composer?

After cloning the repository, install dependencies with make install or uv sync. Configure your MCP servers by editing mcp_servers.json and setting environment variables in .env. Run the application with make run or uv run src/main.py. Alternatively, use make run-docker for a Docker deployment.

Key features of MCP Composer

• Dynamic MCP Server Management with on‑the‑fly activation/deactivation.
• Unified Server-Sent Events (SSE) interface for all managed capabilities.
• Multiple dynamic SSE endpoints for different AI agents or tools.
• Independent interface configuration per SSE endpoint.

Use cases of MCP Composer

• Consolidate multiple MCP servers into a single manageable gateway.
• Provide customized tool/ server combinations to different AI agents.
• Dynamically add or remove MCP servers without restarting the gateway.
• Expose MCP capabilities via a unified SSE interface for workflow tools like n8n.

FAQ from MCP Composer

What is MCP Composer?

MCP Composer is a gateway service that centrally manages all your MCP servers, allowing you to consolidate them and expose independent endpoints with different server/tool combinations for each AI agent or tool.

How do I install and run MCP Composer?

Clone the repository, install dependencies using make install (or uv sync), configure mcp_servers.json and .env, then start with make run or uv run src/main.py. For Docker, run make run-docker.

What are the system requirements?

Python 3.12 or higher and uv (a fast Python package installer/manager).

Can I use Docker to deploy MCP Composer?

Yes, a Docker setup is available. Run make run-docker to start the service with Docker Compose.

How do I configure the MCP servers to manage?

Copy mcp_servers.example.json to mcp_servers.json and edit it to define the MCP servers you want to connect to. Environment variables in .env control host, port, proxy URL, and config path.