CrewAI to MCP

A bridge between CrewAI and Model Context Protocol (MCP) that allows you to expose CrewAI agents and crews as MCP tools.

Overview

This project provides a simple integration between CrewAI and the Model Context Protocol (MCP), enabling you to expose CrewAI agents and crews as MCP-compatible tools. This allows for easy integration of CrewAI's powerful agent orchestration capabilities into any MCP-compatible system.

For the sake of illustration, we use CrewAI's marketing crew example.

Features

• Convert CrewAI classes to MCP servers
• Automatic input schema validation using Pydantic
• Support for SSE (Server-Sent Events) transport
• JSON serialization of CrewAI outputs
• Easy-to-use API for creating MCP tools from CrewAI classes

Installation

uv sync

Configuration

Configure your OpenAI API key and Serper API key in .env

OPENAI_API_KEY=
SERPER_API_KEY=

Usage

Basic Example

from crewai_to_mcp import crewai_to_mcp
from pydantic import BaseModel
from your_crew import YourCrew

# Define your input schema to describe what the inputs to the crew look like.
class InputSchema(BaseModel):
    customer_domain: str
    project_description: str

# Create an MCP server from your CrewAI class
mcp = crewai_to_mcp(
    crewai_class=YourCrew,
    name="YourCrewName",
    description="A description of what your crew does",
    input_schema=InputSchema
)

# Run the server
if __name__ == "__main__":
    mcp.run(transport="sse")

Configuration

The MCP server can be configured using environment variables or a configuration file. See the MCP documentation for more details on configuration options.

Example of Cursor configuration for SSE transport:

{
  "mcpServers": {
    "crew": {
      "url": "http://localhost:8000/sse"
    }
  }
}

Example of Cursor configuration for STDIO transport:

{
    "mcpServers": {
        "crew": {
            "command": "uvx",
            "args": [
                "--from",
                "git+https://github.com/K-Mistele/example-mcp serve_stdio"
            ],
            "env": {
                "OPENAI_API_KEY": "KEY GOES HERE",
                "SERVER_API_KEY": "KEY GOES HERE"
            }
        }
    }
}

API Reference

crewai_to_mcp

def crewai_to_mcp(
    crewai_class: Any,
    name: str,
    description: str,
    input_schema: Type[BaseModel],
) -> FastMCP

Parameters:

• crewai_class: The CrewAI class to convert
• name: The name of the MCP server
• description: The description of the MCP server
• input_schema: A Pydantic model class defining the input schema

Returns:

• A FastMCP server instance

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.