Set up AutoGen
Microsoft’s multi-agent AI framework. Use autogen-ext[mcp] to connect Agentcy tools to your AutoGen agents for autonomous marketing analysis workflows.
- 01
Sign up at app.goagentcy.com and create an API key
- 02
Install AutoGen with MCP: pip install autogen-agentchat 'autogen-ext[openai,mcp]'
- 03
Configure StreamableHTTPServerParams with the Agentcy URL
- 04
Replace YOUR_AGENTCY_API_KEY with your actual key
- 05
Call mcp_server_tools() and pass to an AssistantAgent
path: pip install 'autogen-ext[mcp]'
# Python — AutoGen with MCP
# pip install autogen-agentchat 'autogen-ext[openai,mcp]'
from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient
from autogen_ext.tools.mcp import StreamableHTTPServerParams, mcp_server_tools
server_params = StreamableHTTPServerParams(
url="https://data.goagentcy.com/mcp",
headers={"Authorization": "Bearer YOUR_AGENTCY_API_KEY"},
)
tools = await mcp_server_tools(server_params)
agent = AssistantAgent(
name="marketing_analyst",
model_client=OpenAIChatCompletionClient(model="gpt-4o"),
tools=tools,
)AutoGen supports both mcp_server_tools() (one-shot) and McpWorkbench (persistent connection). Supports stdio, SSE, and Streamable HTTP transports. Works with any AutoGen-compatible LLM. Part of the broader Microsoft Agent Framework ecosystem alongside Semantic Kernel.
“Have a team of agents collaborate on a marketing audit”
“Run a multi-agent analysis of our Google Ads vs organic performance”
“Create an automated reporting pipeline with agent handoffs”
“Analyze competitor strategies across multiple data sources”