Skip to content

Python: fix reasoning model workflow handoff and history serialization #4083

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
TaoChenOSU merged 11 commits into from
Feb 19, 2026
Merged

Python: fix reasoning model workflow handoff and history serialization #4083

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
a5477b9
fix: strip function_call and text_reasoning from cross-agent workflow…
giles17 Feb 19, 2026
3de00bc
fix: reasoning model workflow handoff and history serialization
eavanvalkenburg Feb 19, 2026
b014fce
small improvements in text reasoning
eavanvalkenburg Feb 19, 2026
9ce1f45
refactor: add reset_service_session to AgentExecutorRequest for expli…
eavanvalkenburg Feb 19, 2026
88294f0
comment update
eavanvalkenburg Feb 19, 2026
ab94e00
fixes from feedback
eavanvalkenburg Feb 19, 2026
2aaf01f
fix test
eavanvalkenburg Feb 19, 2026
d9193a2
reverted changes to agent executor
eavanvalkenburg Feb 19, 2026
155e7b9
fix: remove reset_service_session from tau2 runner
eavanvalkenburg Feb 19, 2026
9ff3bd6
two other reverts
eavanvalkenburg Feb 19, 2026
05d336c
fix sample
eavanvalkenburg Feb 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
49 changes: 34 additions & 15 deletions python/packages/core/agent_framework/_tools.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1761,10 +1761,26 @@ def _get_result_hooks_from_stream(stream: Any) -> list[Callable[[Any], Any]]:


def _extract_function_calls(response: ChatResponse) -> list[Content]:
function_results = {it.call_id for it in response.messages[0].contents if it.type == "function_result"}
return [
it for it in response.messages[0].contents if it.type == "function_call" and it.call_id not in function_results
]
function_results = {
item.call_id
for message in response.messages
for item in message.contents
if item.type == "function_result" and item.call_id
}
seen_call_ids: set[str] = set()
function_calls: list[Content] = []
for message in response.messages:
for item in message.contents:
if item.type != "function_call":
continue
if item.call_id and item.call_id in function_results:
continue
if item.call_id and item.call_id in seen_call_ids:
continue
if item.call_id:
seen_call_ids.add(item.call_id)
function_calls.append(item)
return function_calls


def _prepend_fcc_messages(response: ChatResponse, fcc_messages: list[Message]) -> None:
Expand Down Expand Up @@ -1822,27 +1838,22 @@ def _handle_function_call_results(

if had_errors:
errors_in_a_row += 1
if errors_in_a_row >= max_errors:
reached_error_limit = errors_in_a_row >= max_errors
if reached_error_limit:
logger.warning(
"Maximum consecutive function call errors reached (%d). "
"Stopping further function calls for this request.",
max_errors,
)
return {
"action": "stop",
"errors_in_a_row": errors_in_a_row,
"result_message": None,
"update_role": None,
"function_call_results": None,
}
else:
errors_in_a_row = 0
reached_error_limit = False

result_message = Message(role="tool", contents=function_call_results)
response.messages.append(result_message)
fcc_messages.extend(response.messages)
return {
"action": "continue",
"action": "stop" if reached_error_limit else "continue",
"errors_in_a_row": errors_in_a_row,
"result_message": result_message,
"update_role": "tool",
Expand Down Expand Up @@ -2025,6 +2036,7 @@ def get_response(
middleware_pipeline=function_middleware_pipeline,
)
filtered_kwargs = {k: v for k, v in kwargs.items() if k != "session"}

# Make options mutable so we can update conversation_id during function invocation loop
mutable_options: dict[str, Any] = dict(options) if options else {}
# Remove additional_function_arguments from options passed to underlying chat client
Expand Down Expand Up @@ -2090,7 +2102,9 @@ async def _get_response() -> ChatResponse:
if result["action"] == "return":
return response
if result["action"] == "stop":
break
# Error threshold reached: force a final non-tool turn so
# function_call_output items are submitted before exit.
mutable_options["tool_choice"] = "none"
errors_in_a_row = result["errors_in_a_row"]

# When tool_choice is 'required', reset tool_choice after one iteration to avoid infinite loops
Expand Down Expand Up @@ -2157,6 +2171,7 @@ async def _stream() -> AsyncIterable[ChatResponseUpdate]:
)
errors_in_a_row = approval_result["errors_in_a_row"]
if approval_result["action"] == "stop":
mutable_options["tool_choice"] = "none"
return

inner_stream = await _ensure_response_stream(
Expand Down Expand Up @@ -2205,7 +2220,11 @@ async def _stream() -> AsyncIterable[ChatResponseUpdate]:
contents=result["function_call_results"] or [],
role=role,
)
if result["action"] != "continue":
if result["action"] == "stop":
# Error threshold reached: submit collected function_call_output
# items once more with tools disabled.
mutable_options["tool_choice"] = "none"
elif result["action"] != "continue":
return

# When tool_choice is 'required', reset the tool_choice after one iteration to avoid infinite loops
Expand Down
2 changes: 2 additions & 0 deletions python/packages/core/agent_framework/_types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -531,6 +531,7 @@ def from_text(
def from_text_reasoning(
cls: type[ContentT],
*,
id: str | None = None,
text: str | None = None,
protected_data: str | None = None,
annotations: Sequence[Annotation] | None = None,
Expand All @@ -540,6 +541,7 @@ def from_text_reasoning(
"""Create text reasoning content."""
return cls(
"text_reasoning",
id=id,
text=text,
protected_data=protected_data,
annotations=annotations,
Expand Down
10 changes: 5 additions & 5 deletions python/packages/core/agent_framework/_workflows/_agent_executor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -144,10 +144,10 @@ async def from_response(
immediately run the agent to produce a new response.
"""
# Replace cache with full conversation if available, else fall back to agent_response messages.
if prior.full_conversation is not None:
self._cache = list(prior.full_conversation)
else:
self._cache = list(prior.agent_response.messages)
source_messages = (
prior.full_conversation if prior.full_conversation is not None else prior.agent_response.messages
)
self._cache = list(source_messages)
await self._run_agent_and_emit(ctx)

@handler
Expand Down Expand Up @@ -311,7 +311,7 @@ async def _run_agent_and_emit(
# Snapshot current conversation as cache + latest agent outputs.
# Do not append to prior snapshots: callers may provide full-history messages
# in request.messages, and extending would duplicate prior turns.
self._full_conversation = list(self._cache) + (list(response.messages) if response else [])
self._full_conversation = [*self._cache, *(list(response.messages) if response else [])]

if response is None:
# Agent did not complete (e.g., waiting for user input); do not emit response
Expand Down
86 changes: 57 additions & 29 deletions python/packages/core/agent_framework/openai/_responses_client.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -908,11 +908,16 @@ def _prepare_message_for_openai(
"type": "message",
"role": message.role,
}
# Reasoning items are only valid in input when they directly preceded a function_call
# in the same response. Including a reasoning item that preceded a text response
# (i.e. no function_call in the same message) causes an API error:
# "reasoning was provided without its required following item."
has_function_call = any(c.type == "function_call" for c in message.contents)
for content in message.contents:
match content.type:
case "text_reasoning":
# Reasoning items must be sent back as top-level input items
# for reasoning models that require them alongside function_calls
if not has_function_call:
continue # reasoning not followed by a function_call is invalid in input
reasoning = self._prepare_content_for_openai(message.role, content, call_id_to_id) # type: ignore[arg-type]
if reasoning:
all_messages.append(reasoning)
Expand Down Expand Up @@ -961,26 +966,19 @@ def _prepare_content_for_openai(
"text": content.text,
}
case "text_reasoning":
ret: dict[str, Any] = {
"type": "reasoning",
"summary": {
"type": "summary_text",
"text": content.text,
},
}
ret: dict[str, Any] = {"type": "reasoning", "summary": []}
if content.id:
ret["id"] = content.id
props: dict[str, Any] | None = getattr(content, "additional_properties", None)
if props:
if reasoning_id := props.get("reasoning_id"):
ret["id"] = reasoning_id
if status := props.get("status"):
ret["status"] = status
if reasoning_text := props.get("reasoning_text"):
ret["content"] = {
"type": "reasoning_text",
"text": reasoning_text,
}
ret["content"] = [{"type": "reasoning_text", "text": reasoning_text}]
if encrypted_content := props.get("encrypted_content"):
ret["encrypted_content"] = encrypted_content
if content.text:
ret["summary"].append({"type": "summary_text", "text": content.text})
return ret
case "data" | "uri":
if content.has_top_level_media_type("image"):
Expand Down Expand Up @@ -1189,30 +1187,45 @@ def _parse_response_from_openai(
)
)
case "reasoning": # ResponseOutputReasoning
reasoning_id = getattr(item, "id", None)
if hasattr(item, "content") and item.content:
for index, reasoning_content in enumerate(item.content):
added_reasoning = False
if item_content := getattr(item, "content", None):
for index, reasoning_content in enumerate(item_content):
additional_properties: dict[str, Any] = {}
if reasoning_id:
additional_properties["reasoning_id"] = reasoning_id
if hasattr(item, "summary") and item.summary and index < len(item.summary):
additional_properties["summary"] = item.summary[index]
contents.append(
Content.from_text_reasoning(
id=item.id,
text=reasoning_content.text,
raw_representation=reasoning_content,
additional_properties=additional_properties or None,
)
)
if hasattr(item, "summary") and item.summary:
for summary in item.summary:
added_reasoning = True
if item_summary := getattr(item, "summary", None):
for summary in item_summary:
contents.append(
Content.from_text_reasoning(
id=item.id,
text=summary.text,
raw_representation=summary, # type: ignore[arg-type]
additional_properties={"reasoning_id": reasoning_id} if reasoning_id else None,
)
)
added_reasoning = True
if not added_reasoning:
# Reasoning item with no visible text (e.g. encrypted reasoning).
# Always emit an empty marker so co-occurrence detection can be done
additional_properties_empty: dict[str, Any] = {}
if encrypted := getattr(item, "encrypted_content", None):
additional_properties_empty["encrypted_content"] = encrypted
contents.append(
Content.from_text_reasoning(
id=item.id,
text="",
raw_representation=item,
additional_properties=additional_properties_empty or None,
)
)
case "code_interpreter_call": # ResponseOutputCodeInterpreterCall
call_id = getattr(item, "call_id", None) or getattr(item, "id", None)
outputs: list[Content] = []
Expand Down Expand Up @@ -1427,36 +1440,36 @@ def _parse_chunk_from_openai(
case "response.reasoning_text.delta":
contents.append(
Content.from_text_reasoning(
id=event.item_id,
text=event.delta,
raw_representation=event,
additional_properties={"reasoning_id": event.item_id},
)
)
metadata.update(self._get_metadata_from_response(event))
case "response.reasoning_text.done":
contents.append(
Content.from_text_reasoning(
id=event.item_id,
text=event.text,
raw_representation=event,
additional_properties={"reasoning_id": event.item_id},
)
)
metadata.update(self._get_metadata_from_response(event))
case "response.reasoning_summary_text.delta":
contents.append(
Content.from_text_reasoning(
id=event.item_id,
text=event.delta,
raw_representation=event,
additional_properties={"reasoning_id": event.item_id},
)
)
metadata.update(self._get_metadata_from_response(event))
case "response.reasoning_summary_text.done":
contents.append(
Content.from_text_reasoning(
id=event.item_id,
text=event.text,
raw_representation=event,
additional_properties={"reasoning_id": event.item_id},
)
)
metadata.update(self._get_metadata_from_response(event))
Expand Down Expand Up @@ -1630,11 +1643,10 @@ def _parse_chunk_from_openai(
)
case "reasoning": # ResponseOutputReasoning
reasoning_id = getattr(event_item, "id", None)
added_reasoning = False
if hasattr(event_item, "content") and event_item.content:
for index, reasoning_content in enumerate(event_item.content):
additional_properties: dict[str, Any] = {}
if reasoning_id:
additional_properties["reasoning_id"] = reasoning_id
if (
hasattr(event_item, "summary")
and event_item.summary
Expand All @@ -1643,11 +1655,27 @@ def _parse_chunk_from_openai(
additional_properties["summary"] = event_item.summary[index]
contents.append(
Content.from_text_reasoning(
id=reasoning_id or None,
text=reasoning_content.text,
raw_representation=reasoning_content,
additional_properties=additional_properties or None,
)
)
added_reasoning = True
if not added_reasoning:
# Reasoning item with no visible text (e.g. encrypted reasoning).
# Always emit an empty marker so co-occurrence detection can occur.
additional_properties_empty: dict[str, Any] = {}
if encrypted := getattr(event_item, "encrypted_content", None):
additional_properties_empty["encrypted_content"] = encrypted
contents.append(
Content.from_text_reasoning(
id=reasoning_id or None,
text="",
raw_representation=event_item,
additional_properties=additional_properties_empty or None,
)
)
case _:
logger.debug("Unparsed event of type: %s: %s", event.type, event)
case "response.function_call_arguments.delta":
Expand Down
Loading