microsoft/autogen · error · ValueError

Recorded call type mismatch at index {self._record_index}: e

Error message

Recorded call type mismatch at index {self._record_index}: expected 'create', got '{rec.get('mode')}'.

What it means

Raised in replay mode when the record at the current index is not a 'create' record - typically because the current call is create() but the next recorded entry is 'create_stream'. The recorder requires the exact same sequence of call modes as recorded.

Source

Thrown at python/packages/autogen-ext/src/autogen_ext/experimental/task_centric_memory/utils/chat_completion_client_recorder.py:123

            rec: RecordDict = {
                "mode": "create",
                "messages": current_messages,
                "response": response.model_dump(),
                "stream": [],
            }
            self.records.append(rec)
            return response
        elif self.mode == "replay":
            if self._record_index >= len(self.records):
                error_str = "\nNo more recorded turns to check."
                self.logger.error(error_str)
                raise ValueError(error_str)
            rec = self.records[self._record_index]
            if rec.get("mode") != "create":
                error_str = f"\nRecorded call type mismatch at index {self._record_index}: expected 'create', got '{rec.get('mode')}'."
                self.logger.error(error_str)
                raise ValueError(error_str)
            recorded_messages = rec.get("messages")
            if recorded_messages != current_messages:
                error_str = (
                    "\nCurrent message list doesn't match the recorded message list. See the pagelogs for details."
                )
                assert recorded_messages is not None
                self.logger.log_dict_list(recorded_messages, "recorded message list")
                assert current_messages is not None
                self.logger.log_dict_list(current_messages, "current message list")
                self.logger.error(error_str)
                raise ValueError(error_str)
            self._record_index += 1
            self._num_checked_records += 1

            data = rec.get("response")
            # Populate a CreateResult from the data.
            assert data is not None
            result = CreateResult(

View on GitHub (pinned to 027ecf0a37)

Solutions

1. If streaming usage changed, re-record the session with the current code path so modes align.
2. Inspect the session JSON at the reported index to see which mode was recorded and compare with the failing call; fix the divergence (usually an earlier extra or missing call).
3. Keep streaming settings identical between record and replay runs (same agents, same client wrappers).

Example fix

# before
# session recorded with streaming agent, replayed with non-streaming call:
resp = await client.create(messages)  # ValueError: ... expected 'create', got 'create_stream'

# after
# re-record using the same call style you will replay, e.g. streaming:
stream = await client.create_stream(messages)  # or regenerate the session via mode="record"

import json
from pathlib import Path

def next_record_mode(session_file: str, index: int) -> str | None:
    records = json.loads(Path(session_file).read_text())
    return records[index]["mode"] if index < len(records) else None

# before calling create(): assert next_record_mode(p, idx) == "create"

try:
    await client.create(messages)
except ValueError as e:
    if "call type mismatch" in str(e):
        pytest.fail("Streaming usage differs from recording - re-record or use create_stream")
    raise

When it happens

Trigger: The recorded session captured streaming calls (create_stream) but the replayed code now calls non-streaming create(), or an extra/missing call shifted the index so a stream record is read where a create record was expected.

Common situations: Switching a component between streaming and non-streaming usage after recording (changed agent config or model client settings), divergence earlier in the run misaligning indices, library version differences in default streaming behavior.

Related errors

• Failed to load recorded session: '{self.session_file_path}':
• No more recorded turns to check.
• Current message list doesn't match the recorded message list
• Early termination. Only {self._num_checked_records} of the
• No more mock responses available

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/051095f579b87d77. Report an issue: GitHub.