Programmatic Usage(Python)

MCP with AutoGen

Use MCP tools inside AutoGen multi-agent systems

Overview

AutoGen supports MCP through McpWorkbench, which connects to MCP servers and exposes their tools to AssistantAgent instances.


Setup

pip install autogen-agentchat autogen-ext[openai,mcp]

Step 1: Create the MCP Workbench

from autogen_ext.tools.mcp import McpWorkbench, SseServerParams

server_params = SseServerParams(
    url="https://api.mcphero.app/mcp/{server_id}/sse",
    headers={"Authorization": "Bearer <your-api-key>"},
)

Step 2: Create an Agent

from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient

model_client = OpenAIChatCompletionClient(model="gpt-4o")

Step 3: Run the Agent

import asyncio

async def main():
    async with McpWorkbench(server_params) as workbench:
        agent = AssistantAgent(
            name="assistant",
            model_client=model_client,
            workbench=workbench,
            reflect_on_tool_use=True,
        )

        result = await agent.run(
            task="Find customer John Smith and return their last order"
        )
        print(result.messages[-1].content)

    await model_client.close()

asyncio.run(main())

When to Use AutoGen

  • Multi-agent collaboration
  • Tool delegation across agent teams
  • Research or planning workflows

On this page