CrewAI

At the time of writing, MCPAdapt offers integration with CrewAI through its adapter system only. However a PR is underway to integrate directly into CrewAI's framework.

Let's explore how to use MCPAdapt with CrewAI.

Using MCPAdapt CrewAI Adapter

Ensure you have the necessary dependencies installed and set up your OpenAI API key as required by crewAI:

pip install mcpadapt[crewai]

Creating a Simple Echo Server

First, create an MCP server using FastMCP notation:

# echo.py
from mcp.server.fastmcp import FastMCP
from pydantic import Field

# Create server
mcp = FastMCP("Echo Server")


@mcp.tool()
def echo_tool(text: str = Field(description="The text to echo")) -> str:
    """Echo the input text

    Args:
        text (str): The text to echo

    Returns:
        str: The echoed text
    """
    return text

mcp.run()

Using the Echo Server

Here's how to interact with the server using CrewAI:

import os
from crewai import Agent, Crew, Task
from mcp import StdioServerParameters

from mcpadapt.core import MCPAdapt
from mcpadapt.crewai_adapter import CrewAIAdapter

with MCPAdapt(
    StdioServerParameters(command="uv", args=["run", "echo.py"]),
    CrewAIAdapter(),
) as tools:
    # Create an echo agent
    agent = Agent(
        role="Echo Agent",
        goal="Echo messages back to the user",
        backstory="You help echo messages back to users",
        verbose=True,
        tools=[tools[0]],
    )

    # Create a task
    task = Task(
        description="Echo 'Hello, World!'",
        agent=agent,
        expected_output="The echoed message",
    )

    # Create and run the crew
    crew = Crew(agents=[agent], tasks=[task], verbose=True)
    crew.kickoff()

Real-World Example: PubMed MCP Server

In the real-world, you are most likely to run MCP server already built by the community. As an example, here's how to use the PubMed MCP server with CrewAI:

import os
from dotenv import load_dotenv
from crewai import Agent, Crew, Task
from mcp import StdioServerParameters

from mcpadapt.core import MCPAdapt
from mcpadapt.crewai_adapter import CrewAIAdapter

# Load environment variables
load_dotenv()
if not os.environ.get("OPENAI_API_KEY"):
    raise ValueError("OPENAI_API_KEY must be set in your environment variables")

# Initialize MCPAdapt with CrewAI adapter
with MCPAdapt(
    StdioServerParameters(
        command="uvx",
        args=["--quiet", "pubmedmcp@0.1.3"],
        env={"UV_PYTHON": "3.12", **os.environ},
    ),
    CrewAIAdapter(),
) as tools:
    # Create a research agent
    agent = Agent(
        role="Research Agent",
        goal="Find studies about hangover",
        backstory="You help find studies about hangover",
        verbose=True,
        tools=[tools[0]],
    )

    # Create a task
    task = Task(
        description="Find studies about hangover",
        agent=agent,
        expected_output="A list of studies about hangover",
    )

    # Create and run the crew
    crew = Crew(agents=[agent], tasks=[task], verbose=True)
    crew.kickoff()

Important Notes:

• Always use the --quiet flag with uvx to prevent output interference with the stdio transport protocol
• Including **os.environ helps resolve paths in the subprocess but consider security implications of sending your environment variables to the MCP server
• Remote MCP servers are supported via Server Sent Events (SSE). See the quickstart SSE guide
• Make sure to set your OPENAI_API_KEY in your environment variables or .env file

Full Working Code Example

You can find a fully working script of this example here