How to Intermediate · 3 min read

How to run MCP server with SSE

Q: How to run MCP server with SSE

Run an MCP server with SSE transport by using the official mcp Python SDK's stdio_server with SSE enabled. Import Server and stdio_server from mcp.server, then start the server with stdio_server() configured for SSE. This enables event-stream communication for AI agents.

Quick answer

Run an MCP server with SSE transport by using the official mcp Python SDK's stdio_server with SSE enabled. Import Server and stdio_server from mcp.server, then start the server with stdio_server() configured for SSE. This enables event-stream communication for AI agents.

PREREQUISITES

Python 3.8+
pip install mcp
Basic knowledge of Python async programming

Setup

Install the official MCP Python SDK and ensure Python 3.8 or higher is installed. The MCP SDK provides the Server class and stdio_server helper to run MCP servers with different transports including SSE.

bash

pip install mcp

Step by step

This example shows how to run a basic MCP server using SSE transport. The server listens for MCP messages over Server-Sent Events and responds accordingly.

python

from mcp.server import Server
from mcp.server.stdio import stdio_server

# Define a simple MCP server handler
class MyMCPServer(Server):
    async def handle_message(self, message):
        # Echo back the received message content
        response = {
            "id": message.get("id"),
            "type": "response",
            "data": f"Received: {message.get('data', '')}"
        }
        return response

# Run the MCP server with SSE transport
if __name__ == "__main__":
    # stdio_server runs the server using SSE by default
    stdio_server(MyMCPServer())

output

Running MCP server with SSE transport...
(Server listens for incoming MCP messages and responds)

Common variations

Async server: The MCP server handler methods are async, allowing concurrency.
Different transports: Besides SSE, MCP supports stdio transport; use stdio_server() for SSE.
Custom message handling: Override handle_message to implement your logic.

Troubleshooting

If the server does not respond, ensure the client supports SSE and the connection is not blocked by proxies.
Check Python version is 3.8 or higher to support async features.
Use logging inside handle_message to debug message flow.

✅

Key Takeaways

Use the official mcp Python SDK to run MCP servers with SSE transport easily.
Implement async handle_message in your Server subclass for custom logic.
Run the server with stdio_server(MyMCPServer()) to enable SSE communication.
Ensure clients support SSE and Python version is 3.8+ for async compatibility.

Verified 2026-04

Verify ↗