Incorrect typing hint for FastMCP.call_tool

[de-code]

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

The signature for FastMCP.call_tool:

 async def call_tool(self, name: str, arguments: dict[str, Any]) -> Sequence[ContentBlock] | dict[str, Any]:

i.e. it says the return type is either a sequence of ContentBlock or a dict.

However, it seems to be more like either just a sequence of ContentBlock or a tuple of: sequence of ContentBlock and a dict:

Sequence[ContentBlock] | tuple[Sequence[ContentBlock], dict[str, Any]]

That is because it always passes convert_result=True to self._tool_manager.call_tool (it would be nice if that was exposed as a parameter). And the convert_result result function (in mcp.server.fastmcp.utilities.func_metadata) is implemented like that but doesn't have any useful type hints.

This has changed after version 1.9.4 (i.e. in 1.9.4 it was still just a sequence of ContentBlock).

Example Code

from mcp.server.fastmcp import FastMCP


mcp = FastMCP(name='Example MCP Server')


@mcp.tool()
async def add_numbers(a: int, b: int) -> int:
 """Add two numbers."""
 return a + b

...

# Within an async context call:

result = await mcp.call_tool(
 "add_numbers",
 arguments={"a": 1, "b": 2}
)
print('result:', result)

# Output:
result: ([TextContent(type='text', text='3', annotations=None, meta=None)], {'result': 3})

Python & MCP Python SDK

1.12.4

[maxisbey]

Thanks for the report. Closing this as it's been fixed in v2, which is now released (see #2816). If you still run into this on v2, feel free to reopen.

_{AI Disclaimer}