Progress notifications cause server to hang on stdio transport

[abhishekgahlot2]

Initial Checks

• I confirm that I'm using the latest version of MCP Python SDK
• I confirm that I searched for my issue in https://github.com/modelcontextprotocol/python-sdk/issues before opening this issue

Description

When using the low-level Server API (mcp.server.lowlevel.Server) with stdio transport, sending progress notifications during request handling causes the server to hang indefinitely and never send the final response.

What I am seeing:

• Server successfully sends progress notification via write_stream.send()
• Server handler returns CallToolResult
• The final response is never sent to the client
• Client times out waiting for response
• Server remains blocked and time out eventually

Example Code

import asyncio
from mcp.server.lowlevel import Server
from mcp.server.stdio import stdio_server
from mcp.types import *

server = Server("progress-bug-demo")

@server.list_tools()
async def handle_list_tools() -> list[Tool]:
 return [Tool(
 name="test_tool",
 description="Demonstrates the bug",
 inputSchema={"type": "object", "properties": {}}
 )]

@server.call_tool()
async def handle_call_tool(name: str, arguments: dict) -> list[TextContent]:
 ctx = server.request_context
 progress_token = ctx.meta.progressToken if ctx.meta else None
 
 if progress_token:
 # This causes the server to hang
 await ctx.session.send_progress_notification(
 progress_token=progress_token,
 progress=0.5,
 total=1.0,
 message="Working..."
 )
 
 # This response is never sent due to the hang
 return [TextContent(type="text", text="Task completed!")]

async def main():
 async with stdio_server() as (read_stream, write_stream):
 await server.run(
 read_stream,
 write_stream,
 server.create_initialization_options()
 )

if __name__ == "__main__":
 asyncio.run(main())

Python & MCP Python SDK

MCP: 1.10.1
Python: 3.12.6

From deeper debugging looks like buffer size is the problem : https://github.com/modelcontextprotocol/python-sdk/blob/main/src/mcp/server/stdio.py#L57

Have to test locally if increasing it will most likely fix the issue and if its causing more problems than just fixing this.

[abhishekgahlot2]

Actually using buffered streams seems to fix this problem.

[ihrpr]

@abhishekgahlot2 which client are you using for testing?

[abhishekgahlot2]

@ihrpr I am using a custom Python MCP client that I built for agents.

However looking more into my custom client i see sending progress via

# cause deadlock
await self.write_stream.send({...})

# sdk works well here
await ctx.session.send_progress_notification(...)