mcp-copilot

Overview

MCP-Copilot acts as a router for multiple MCP servers, allowing users to access thousands of tools through intelligent routing. It provides three main tools: router-servers to find relevant servers, route-tools to discover appropriate tools across all servers, and execute-tool to run specific tools. This architecture prevents LLMs from being overwhelmed by exposing all available tools at once.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• router-serversquery (string, required), top_k (integer, optional)

  Route user query to appropriate MCP servers.

  Try:Find servers for data analysisTry:Show me servers related to file management
• route-toolsquery (string, required), top_k (integer, optional)

  Route user query to appropriate MCP tools across all servers.

  Try:Find tools for creating GitHub issuesTry:Show me tools for database queries
• execute-toolserver_name (string, required), tool_name (string, required), params (object, optional)

  Execute a specific tool on a specific server based on previous routing results.

  Try:Run the 'query_postgres' tool on the database serverTry:Execute the 'create_github_issue' tool with parameters

Comparable tools

Installation

uvx mcp-server-copilot --config ~/.config/mcp-server-copilot/config.json

pip install mcp-server-copilot

python -m mcp_server_copilot

{
  "mcpServers": {
    "copilot": {
      "command": "uvx",
      "args": ["mcp-server-copilot", "--config", "~/.config/mcp-server-copilot/config.json"]
    }
  }
}

FAQ

How does MCP-Copilot handle routing between servers?

It uses a query-based routing system that matches user requests to the most relevant MCP servers and tools without exposing all available options to the LLM at once.

What is the maximum number of MCP servers it can handle?

The documentation mentions scaling to 1000+ MCP servers, though the actual limit would depend on system resources and configuration.

Compare mcp-copilot with