Three changes to prevent pytest-xdist workers from crashing during
Azure Functions integration tests:
1. Add `start_new_session=True` to subprocess on Linux so signals
(e.g. from test-timeout) cannot propagate between the func host
and the xdist worker process.
2. Add an overall 100-second budget to the fixture setup loop so
the retry logic never exceeds the 120-second test timeout. When
pytest-timeout's thread method fires during fixture setup and the
thread doesn't respond, it calls os._exit() which kills the
xdist worker – this is the root cause of the "Not properly
terminated" crashes.
3. Remove the `UV_PYTHON: "3.10"` workaround from both workflow
files so integration tests actually run on Python 3.13.
Co-authored-by: larohra <41490930+larohra@users.noreply.github.com>
* small updates and improvements in the azure AISearch provider
* Fix mypy errors and embedding function test
- Use separate variable for embeddings result to avoid mypy type reassignment error
- Fix test_vectorized_query_with_embedding_function: use real async function
instead of AsyncMock which falsely matches SupportsGetEmbeddings protocol
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fixes from feedback
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix thread corruption when max_iterations exhausted (#1366)
When the function invocation loop exhausts max_iterations while the model
keeps requesting tools, the failsafe code path (calling the model with
tool_choice='none' and prepending fcc_messages) was unreachable because
'if response is not None: return response' short-circuited before it.
The fix removes the premature return so the failsafe always runs after
loop exhaustion, making a final model call with tool_choice='none' to
produce a clean text answer and prepending accumulated fcc_messages from
prior iterations. This matches the existing pattern used by the error
threshold and max_function_calls paths.
Also unskips test_max_iterations_limit and test_streaming_max_iterations_limit
which were previously skipped with 'needs investigation in unified API'.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add fix report for issue #1366
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix ruff formatting in _tools.py and test_issue_1366_thread_corruption.py
Apply ruff format to fix multi-line string concatenation and function call
formatting issues flagged by the linter.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add quality review for issue #1366 fix
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Remove temporary investigation docs.
* Address PR review: explicit enabled check in log condition, clarify mock behavior in test
- Add explicit function_invocation_configuration['enabled'] check to the
'Maximum iterations reached' log condition in both non-streaming and
streaming paths, making intent clearer when function invocation is disabled.
- Add comment in test_thread_safe_after_max_iterations_with_agent explaining
that the failsafe response (tool_choice='none') is provided automatically
by the mock client, not from run_responses.
* Blend fix and tests into project without issue-specific callouts
- Remove issue #1366 references from _tools.py comments
- Move regression tests from standalone test_issue_1366_thread_corruption.py
into test_function_invocation_logic.py alongside existing max_iterations tests
- Clean up test docstrings to describe behavior generically
- Delete the standalone issue-specific test file
---------
Co-authored-by: alliscode <bentho@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: prevent doubled tool_call arguments in MESSAGES_SNAPSHOT
When streaming with client-side tools, some providers send a full-
arguments replay after the streaming deltas complete. The `_emit_tool_call`
function unconditionally appends every arguments delta to the internal
`flow.tool_calls_by_id` tracking dictionary via `+=`. When the replay
contains the exact same complete arguments string that was already
accumulated from prior deltas, the arguments get doubled (e.g.,
`{"todoText":"buy groceries"}{"todoText":"buy groceries"}`).
This causes `MESSAGES_SNAPSHOT` events to contain invalid doubled JSON in
`tool_calls[].function.arguments`, breaking any client or middleware that
relies on snapshots for state reconstruction.
The fix adds a guard (mirroring the existing duplicate guard in
`_emit_text`) that detects when the incoming delta exactly equals the
already-accumulated arguments string, indicating a full-arguments replay
rather than an incremental delta. In this case the append is skipped,
preventing the doubling.
The `ToolCallArgsEvent` deltas are still emitted correctly for real-time
streaming — only the internal snapshot accumulator is guarded.
Fixes#4194
* fix: move duplicate check before event emission + add test
Address Copilot review feedback:
1. Move duplicate full-arguments replay detection BEFORE emitting
ToolCallArgsEvent, for consistency with _emit_text() which returns
early without emitting any events on replay detection.
2. Add test_emit_tool_call_skips_duplicate_full_arguments_replay() to
verify the duplicate detection behavior for tool call arguments,
matching the existing test pattern for text content.
* updated integration tests and guidance
* fixed merge test
* updated integration tests
* fix: remove duplicate --dist loadfile flag from pytest-xdist config
Only one --dist mode can be active at a time; the second value silently
overrides the first. Keep --dist worksteal (dynamic load balancing) and
remove the redundant --dist loadfile from all workflow files and
pyproject.toml configs.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* docs: add keep-in-sync notes for merge and integration test workflows
Both python-merge-tests.yml and python-integration-tests.yml share the
same parallel job structure. Added sync reminders in workflow file
comments, the python-testing SKILL.md, and CODING_STANDARD.md.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* refactor: remove RUN_INTEGRATION_TESTS flag
Integration test gating now uses two mechanisms:
- `@pytest.mark.integration` for test selection via `-m` filtering
- `skip_if_*_disabled` for credential/service availability checks
The RUN_INTEGRATION_TESTS env var was redundant since the marker handles
selection and the skip decorators already check for actual credentials.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: sync missing env vars from merge-tests to integration-tests
Add OPENAI_EMBEDDINGS_MODEL_ID and AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME
to python-integration-tests.yml to match python-merge-tests.yml.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: remove remaining RUN_INTEGRATION_TESTS from embedding tests and docs
Missed test_openai_embedding_client.py and vector-stores README in the
earlier cleanup.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* set functions tests to 3.10
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* feat(python): Add embedding abstractions and OpenAI implementation (Phase 1)
This PR contains two parts:
1. **Overall migration plan** for porting vector stores and embeddings from
Semantic Kernel to Agent Framework (docs/features/vector-stores-and-embeddings/README.md)
covering all 10 phases from core abstractions through connectors and TextSearch.
2. **Phase 1 implementation** — core embedding abstractions and OpenAI/Azure OpenAI
embedding clients:
Core types (_types.py):
- EmbeddingGenerationOptions TypedDict (total=False)
- Embedding[EmbeddingT] generic class with model_id, dimensions, created_at
- GeneratedEmbeddings[EmbeddingT, EmbeddingOptionsT] list container with options, usage
- EmbeddingInputT (default str) and EmbeddingT (default list[float]) TypeVars
Protocol + base class (_clients.py):
- SupportsGetEmbeddings protocol — Generic[EmbeddingInputT, EmbeddingT, OptionsContraT]
- BaseEmbeddingClient ABC — Generic[EmbeddingInputT, EmbeddingT, OptionsCoT]
Telemetry (observability.py):
- EmbeddingTelemetryLayer with gen_ai.operation.name = "embeddings"
OpenAI implementation (openai/_embedding_client.py):
- RawOpenAIEmbeddingClient, OpenAIEmbeddingClient, OpenAIEmbeddingOptions
- Uses _ensure_client() factory pattern
Azure OpenAI implementation (azure/_embedding_client.py):
- AzureOpenAIEmbeddingClient following AzureOpenAIChatClient pattern
- Supports API key, Entra ID credentials, env var configuration
Tests:
- 47 unit tests for types, protocol, base class, OpenAI, and Azure clients
- 6 integration tests (gated behind RUN_INTEGRATION_TESTS + credentials)
Samples:
- samples/02-agents/embeddings/openai_embeddings.py
- samples/02-agents/embeddings/azure_openai_embeddings.py
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Add AzureOpenAIEmbeddingClient to azure __init__.pyi stub
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* ci: Add embedding env vars to Python integration tests
Map OPENAI_EMBEDDING_MODEL_ID and AZURE_OPENAI_EMBEDDING_DEPLOYMENT_NAME
from GitHub vars to the integration test environment.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Handle base64 encoding_format in OpenAI embedding client
When encoding_format='base64' is used, the OpenAI API returns base64-encoded
floats instead of a JSON array. Decode these automatically to list[float]
so the return type stays consistent regardless of encoding format.
Also adds a unit test for base64 decoding and fixes minor docstring/import issues.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Only record INPUT_TOKENS for embedding telemetry
Embeddings have no output/completion tokens. Remove OUTPUT_TOKENS recording
which was double-counting prompt_tokens via the total_tokens fallback.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Resolve mypy variance error and lint warning
Use contravariant/covariant TypeVars for SupportsGetEmbeddings Protocol.
Combine nested if into single statement in telemetry layer.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Make EmbeddingCoT invariant for mypy compatibility
GeneratedEmbeddings is invariant in its type param, so the Protocol
TypeVar cannot be covariant.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Address PR review - empty values guard, service_url for telemetry
- Add early return for empty values in get_embeddings to avoid unnecessary API calls
- Add service_url() method to RawOpenAIEmbeddingClient for proper telemetry endpoint reporting
- Add test for empty values behavior
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Python: Fix OpenAI chat client compatibility with third-party endpoints and OTel 0.4.14 (#4161)
* Fix system message content sent as list instead of string
Some OpenAI-compatible endpoints (e.g. NVIDIA NIM) reject system messages
when content is a list of content parts. This change flattens system and
developer message content to a plain string in the Chat Completions client.
Fixes https://github.com/microsoft/agent-framework/issues/1407
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix compatibility with opentelemetry-semantic-conventions-ai 0.4.14
Version 0.4.14 removed several LLM_* attributes from SpanAttributes
(LLM_SYSTEM, LLM_REQUEST_MODEL, LLM_RESPONSE_MODEL, LLM_REQUEST_MAX_TOKENS,
LLM_REQUEST_TEMPERATURE, LLM_REQUEST_TOP_P, LLM_TOKEN_TYPE).
Move these to the OtelAttr enum with their well-known gen_ai.* string values
and update all references in observability.py and tests.
Fixes https://github.com/microsoft/agent-framework/issues/4160
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Flatten text-only message content to string for all roles
Extend the system/developer fix to all message roles. Text-only content
lists are now post-processed into plain strings, while multimodal content
(text + images/audio) remains as a list. This fixes compatibility with
OpenAI-like endpoints that cannot deserialize list content (e.g. Foundry
Local's Neutron backend).
Partially fixes https://github.com/microsoft/agent-framework/issues/4084
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix streaming text lost when usage data in same chunk
Some providers (e.g. Gemini) include both usage data and text content
in the same streaming chunk. The early return on chunk.usage caused
text and tool call parsing to be skipped entirely. Remove the early
return and process usage alongside text/tool calls.
Fixes https://github.com/microsoft/agent-framework/issues/3434
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix mypy errors in _chat_client.py
Rename shadowed variable 'args' in system/developer branch to 'sys_args'
and rename loop variable 'content' to 'msg_content' to avoid type conflict.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* reorder imports
* fix: Use OtelAttr.REQUEST_MODEL instead of removed SpanAttributes.LLM_REQUEST_MODEL
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* docs: Add score_threshold to vector store plan
Reference SK .NET PR #13501 for score threshold filtering semantics.
Include score_threshold in SearchOptions from Phase 3.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* docs: Add reference to roji's SK .NET MEVD work for SQL connectors
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: Clear env vars in construction tests to avoid CI leakage
Tests for missing API key / model ID now use monkeypatch.delenv to ensure
env vars from the integration test environment don't prevent the expected
ValueError from being raised.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Python: Enhance Azure AI Search citations with document URLs in Foundry V2 (Responses API)
Override _parse_response_from_openai and _parse_chunk_from_openai in
RawAzureAIClient to extract get_urls from azure_ai_search_call_output
items and enrich url_citation annotations with document-specific URLs.
- Non-streaming: first pass collects get_urls, post-processes annotations
- Streaming: captures search output state, enriches url_citation events
(also handles url_citation annotation type not handled by base class)
- Updated V2 sample to demonstrate citation URL extraction
- Added 14 unit tests covering extraction, enrichment, and edge cases
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* refactor: rework search citation enrichment to override _inner_get_response
- Remove all direct openai/pydantic imports from _client.py
- Override _inner_get_response instead of _parse_response_from_openai/_parse_chunk_from_openai
- Use closure-local state for streaming instead of instance-level _streaming_search_get_urls
- Add _build_url_citation_content helper for streaming url_citation handling
- Fix mypy errors by using str(value or '') for Annotation TypedDict fields
- Fix docstring to say 'citation' instead of 'url_citation'
- Update tests to match new approach
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: handle streaming search citations from output_item.done events
The azure_ai_search_call_output item only has populated output data
(including get_urls) in the response.output_item.done event, not in
the response.output_item.added event. Also removed the search_get_urls
guard on url_citation handling so annotations are always produced even
if get_urls haven't been captured yet.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* addressed comments
* refactor: address PR review - eliminate type: ignore[assignment] pattern
Call super()._inner_get_response() independently in each branch instead
of once at the top with union type reassignment. Non-streaming uses
two-arg super() in the closure; streaming uses cast() for type narrowing.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* refactor: remove defensive patterns per PR review
- Replace all getattr() with direct attribute access
- Remove cast() for streaming branch, use type: ignore[assignment]
- Simplify _build_url_citation_content to use dict access directly
- Simplify _extract_azure_search_urls to use item.type/item.output
- Handle empty list output from streaming 'added' events
- Update tests to match actual runtime types (objects, not dicts)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* mypy fix
* small fixes
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add max_function_calls to FunctionInvocationConfiguration (#2329)
Add a new per-request max_function_calls setting to FunctionInvocationConfiguration
that limits the total number of individual function invocations across all iterations
within a single get_response call. This complements max_iterations (which limits LLM
roundtrips) by providing a hard cap on actual tool executions regardless of parallelism.
- Add max_function_calls field to FunctionInvocationConfiguration (default: None/unlimited)
- Track cumulative function call count in both streaming and non-streaming tool loops
- Force tool_choice='none' when the limit is reached
- Add validation in normalize_function_invocation_configuration
- Improve docstrings for FunctionInvocationConfiguration, FunctionTool, and @tool
to clarify semantics of max_iterations vs max_function_calls vs max_invocations
- Add tests for parallel calls, single calls, unlimited mode, and config validation
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add sample for controlling total tool executions
Showcases all three mechanisms for limiting tool executions:
1. max_iterations — caps LLM roundtrips
2. max_function_calls — caps total individual function invocations per request
3. max_invocations — lifetime cap on a specific tool instance
Plus a combined scenario demonstrating defense in depth.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Suppress ruff E305/fmt in hosting sample to preserve XML doc tags
The XML snippet tags (# <create_agent> / # </create_agent>) are used for
docs extraction and must stay adjacent to the code they wrap. Both ruff
check (E305) and ruff format add blank lines after the function definition,
pushing the closing tag away. Suppress with ruff: noqa: E305 and fmt: off.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add per-agent tool wrapping scenario to control_total_tool_executions sample
Show that wrapping the same callable with @tool multiple times creates
independent FunctionTool instances with separate invocation counters,
enabling per-agent max_invocations budgets for shared functions.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Clarify max_function_calls is a best-effort limit
The limit is checked after each batch of parallel calls completes, so the
current batch always runs to completion even if it overshoots the limit.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Address PR review: fix docstring reference, clarify best-effort in sample
- Fix malformed Sphinx :attr: role in FunctionTool docstring — use plain
backtick reference instead
- Update sample to say 'best-effort cap' instead of 'hard cap' for
max_function_calls, noting it's checked between iterations
- Parametrize pattern is correct (fixture override, matching existing tests)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* clarify max_invocations limits
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix structured_output propagation in ClaudeAgent
Capture structured_output from ResultMessage in _get_stream() and
propagate it to AgentResponse.value via a custom finalizer. Previously
structured_output was silently discarded, making output_format unusable.
Fixes#4095
* Address review feedback: use value parameter instead of private properties
- Extend AgentResponse.from_updates() to accept optional value parameter
- Remove structured_output yield from _get_stream()
- Update _finalize_response() to pass value via public API
- Update streaming test to use get_final_response()
* Fix mypy errors: add value parameter to from_updates overloads
Add value parameter to both @overload signatures of
AgentResponse.from_updates() so mypy recognizes the argument.
---------
Co-authored-by: Amit Mukherjee <amimukherjee@microsoft.com>
Co-authored-by: Dmytro Struk <13853051+dmytrostruk@users.noreply.github.com>
* fix Workflow.as_agent() streaming regression in ag-ui
* Address PR feedback
* workflows wip
* wip
* wip
* Workflow AG-UI demo
* Fixes for handoff workflow demo
* Fixes to workflows support in AG-UI
* Fixes
* Add headers to some demo files
* Fix comment
* Fixes for store
* Make _input_schema lazy-loaded
* fix mypy
* revert session change to handoff only for now
---------
Co-authored-by: Eduard van Valkenburg <eavanvalkenburg@users.noreply.github.com>
* Fix system message content sent as list instead of string
Some OpenAI-compatible endpoints (e.g. NVIDIA NIM) reject system messages
when content is a list of content parts. This change flattens system and
developer message content to a plain string in the Chat Completions client.
Fixes https://github.com/microsoft/agent-framework/issues/1407
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix compatibility with opentelemetry-semantic-conventions-ai 0.4.14
Version 0.4.14 removed several LLM_* attributes from SpanAttributes
(LLM_SYSTEM, LLM_REQUEST_MODEL, LLM_RESPONSE_MODEL, LLM_REQUEST_MAX_TOKENS,
LLM_REQUEST_TEMPERATURE, LLM_REQUEST_TOP_P, LLM_TOKEN_TYPE).
Move these to the OtelAttr enum with their well-known gen_ai.* string values
and update all references in observability.py and tests.
Fixes https://github.com/microsoft/agent-framework/issues/4160
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Flatten text-only message content to string for all roles
Extend the system/developer fix to all message roles. Text-only content
lists are now post-processed into plain strings, while multimodal content
(text + images/audio) remains as a list. This fixes compatibility with
OpenAI-like endpoints that cannot deserialize list content (e.g. Foundry
Local's Neutron backend).
Partially fixes https://github.com/microsoft/agent-framework/issues/4084
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix streaming text lost when usage data in same chunk
Some providers (e.g. Gemini) include both usage data and text content
in the same streaming chunk. The early return on chunk.usage caused
text and tool call parsing to be skipped entirely. Remove the early
return and process usage alongside text/tool calls.
Fixes https://github.com/microsoft/agent-framework/issues/3434
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix mypy errors in _chat_client.py
Rename shadowed variable 'args' in system/developer branch to 'sys_args'
and rename loop variable 'content' to 'msg_content' to avoid type conflict.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: strip function_call and text_reasoning from cross-agent workflow handoff
When a reasoning model (e.g. gpt-5-mini) runs as Agent 1 in a workflow, its
response includes text_reasoning items (with server-scoped IDs like rs_XXXX)
and function_call items. Forwarding these to Agent 2 in a fresh conversation
caused API errors because the reasoning/call IDs are scoped to the original
stored response context.
Changes:
- Strip 'function_call', 'text_reasoning', 'function_approval_request', and
'function_approval_response' from handoff messages in _agent_executor.py
- Keep 'function_result' so the actual tool output content is preserved for
the next agent's context
- Update unit tests to reflect that function_result messages survive handoff
(messages grow from 2→3: user, tool(result), assistant(summary))
- Fix incorrect test assertions in test_function_invocation_stop_clears_*
that assumed the client layer updates session.service_session_id
- Also fixed _extract_function_calls to search all messages with call_id
deduplication, and the error-limit stop path to submit function_call_output
items before halting (via tool_choice=none cleanup call)
Relates to: https://github.com/microsoft/agent-framework/issues/4047
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: reasoning model workflow handoff and history serialization
Fixes multiple related issues when using reasoning models (gpt-5-mini,
gpt-5.2) in multi-agent workflows that chain agents via from_response
or replay full conversation history via AgentExecutorRequest.
## Reasoning items always emitted on output_item.added
When a reasoning model produces encrypted or hidden reasoning (no
visible text), the Responses API still fires a reasoning output item
without any reasoning_text.delta events. Previously no text_reasoning
Content was emitted in that case, making it invisible to downstream
logic. Both the non-streaming (_parse_response_from_openai) and
streaming (output_item.added) paths now always emit at least one
text_reasoning Content — with empty text if no content is available —
so co-occurrence detection and serialization guards work reliably.
## Reasoning items only serialized when paired with a function_call
The Responses API only accepts reasoning items in input when they
directly preceded a function_call in the original response. Sending a
reasoning item that preceded a text response (no tool call) causes:
"reasoning was provided without its required following item"
_prepare_message_for_openai now checks has_function_call per message
and skips text_reasoning serialization when there is no accompanying
function_call.
## summary field is an array, not an object
The reasoning item summary field sent to the Responses API must be an
array of objects ([{"type": "summary_text", "text": ...}]), not a
single object. Fixed _prepare_content_for_openai accordingly.
## service_session_id cleared when explicit history is provided
When a workflow coordinator replays a full conversation (including
function calls from a previous agent run) back to an executor via
AgentExecutorRequest or from_response, the executor's session still
held a service_session_id (previous_response_id) from the prior run.
The API then received the same function-call items twice — once from
previous_response_id (server-stored) and once from the explicit input —
causing: "Duplicate item found with id fc_...".
AgentExecutor.run (when should_respond=True) and from_response now
reset self._session.service_session_id = None before running so that
explicit input is the sole source of conversation context.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* small improvements in text reasoning
* refactor: add reset_service_session to AgentExecutorRequest for explicit history replay
Replace the implicit 'always clear service_session_id when should_respond=True'
with an explicit opt-in field on AgentExecutorRequest.
The old approach used should_respond=True as a proxy for 'full history replay',
but that conflates two distinct intents:
- Orchestrations group chat sends should_respond=True with an empty/single-message
list (not a full replay) — unnecessarily clearing service_session_id.
- HITL / feedback coordinators send the full prior conversation and truly need
a fresh service session ID to avoid duplicate-item API errors.
Changes:
- Add AgentExecutorRequest.reset_service_session: bool = False
- AgentExecutor.run only clears service_session_id when this flag is True
- AgentExecutor.from_response unchanged (always clears; always full conversation)
- Set reset_service_session=True in all full-history-replay call sites:
agents_with_HITL.py, azure_chat_agents_tool_calls_with_feedback.py,
autogen-migration round-robin coordinator, tau2 runner
- Update _FullHistoryReplayCoordinator test helper to pass the flag
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* comment update
* fixes from feedback
* fix test
* reverted changes to agent executor
* fix: remove reset_service_session from tau2 runner
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* two other reverts
* fix sample
---------
Co-authored-by: Giles Odigwe <79032838+giles17@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Python: improve .env precedence and observability samples
- Switch load_settings to explicit precedence: overrides -> explicit .env -> environment -> defaults\n- Raise when env_file_path is provided but missing\n- Update settings docs and tests for new behavior\n- Refresh observability samples and README guidance for env loading options\n\nCloses #3864\n\nCo-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fixed some imports
* Fix load_settings CI regressions
Allow explicit env_file_path values that exist but are not regular files (for example /dev/null) by checking path existence before dotenv parsing, and restore a dict accumulator with typed return cast to satisfy mypy.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Avoid implicit dotenv in observability
Only load dotenv in observability helpers when env_file_path is explicitly provided, and remove test os.devnull workarounds that are no longer necessary.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix streaming branch in weather override middleware sample
The streaming branch of weather_override_middleware only prefixed the
original weather data via a transform hook instead of replacing the
content with the 'perfect weather' override like the non-streaming
branch does. Replace with a new ResponseStream that yields the override
content as ChatResponseUpdate chunks.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fixed exception handling middleware sample
* Fixed runtime context delegation middleware example
* Fixed multimodal input examples
* Small update
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Add workflow support for Azure Functions
* fix compatability with latest framework changes and add integration tests
* refactor code
* remove white space
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* align help text with actual port used
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* replace instance id with a place holder
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* remove unused import
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* remove redundant typing import and fix SIM115
* fix latest breaking changes
* fix mypy issues
* clean up imports
* define source marker strings as constants
* fix json module name
* refactor _extract_message_content_from_dict
* refactor serialization
* add helper method for error response construction and remove _extract_message_content_from_dict since it is not needed
* use strict tpe checking for edges
* change how duplicate agent registrations are handled
* cancel approval_task on HITL timeout
* update docstring
* fix: align azurefunctions package with core API changes after rebase
- State.import_state/export_state are now sync (removed await)
- Add State.commit() before export_state() in activity execution
- Rename executor parameter shared_state -> state
- Rename ctx.set_shared_state/get_shared_state -> set_state/get_state (sync)
- WorkflowBuilder now takes start_executor as constructor kwarg
- Update WorkflowOutputEvent -> WorkflowEvent with type='output'
- Update RequestInfoEvent -> WorkflowEvent[Any]
- Update SharedState -> State in test imports
- Update duplicate agent name tests to match new warning behavior
- Update sample README API references
* fix sample check errors
* fix mypy issues
* fix trailing white spaces
* fix test imports
* feat: add durable workflow samples and adapt to main branch changes
- Add workflow samples 09-12 to 04-hosting/azure_functions/
- Adapt to ChatMessage -> Message rename from main
- Adapt to pickle-based checkpoint encoding from main
- Simplify _serialization.py to delegate to core encode/decode
- Fix Message -> WorkflowMessage disambiguation in _context.py
- Remove non-existent _checkpoint_summary import
* fix: update create_checkpoint signature to match superclass
* fix: correct relative link in HITL sample README
* fix: resolve import breakage after rebase (State, DurableAgentThread, get_logger)
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Dmytro Struk <13853051+dmytrostruk@users.noreply.github.com>
* feat: Inject OpenTelemetry trace context into MCP requests and update documentation
* Update python/samples/getting_started/observability/README.md
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Update python/packages/core/tests/core/test_mcp.py
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* refactor: move opentelemetry import to module level
OpenTelemetry is a hard dependency of agent-framework-core (per
pyproject.toml), so the try/except ImportError guard was dead code.
Move the import to the top of the file to fail fast on missing
dependencies instead of silently hiding installation issues.
---------
Co-authored-by: Pete Roden <Pete.Roden@microsoft.com>
Co-authored-by: Mark Wallace <127216156+markwallace-microsoft@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Fix#3600: Pass JSON schemas through without Pydantic conversion
This change optimizes FunctionTool and MCP flows by passing JSON schemas
directly to providers without converting them to Pydantic models first.
Key changes:
- Store JSON schema as-is when supplied to FunctionTool
- Skip Pydantic model_validate for schema-supplied tools in invoke()
- Return MCP tool schemas directly without conversion
- Add comprehensive tests for schema passthrough behavior
Performance benefits:
- Eliminates expensive Pydantic model creation for supplied schemas
- Preserves exact schema structure (additionalProperties, custom fields, etc.)
- Reduces memory overhead and initialization time
Maintains backward compatibility:
- Function signature inference still uses Pydantic models
- Explicit Pydantic models passed as input_model work as before
- All existing tests pass
* Fix schema passthrough validation and remove helper
* Simplify FunctionTool without generic model dependency
* Fix FunctionTool typing fallout in 3600
* Remove FunctionTool[Any] compatibility shim
* Use serializable kwargs in OTEL tool args
* Python: Replace wildcard imports with explicit imports
- Replace all 'from ... import *' with explicit symbol imports
- Add __all__ declarations to namespace packages for re-exports
- Update CODING_STANDARD.md to prohibit wildcard imports
- Maintain exported API and preserve all functionality
fixes#3605
* Refine wildcard guidance example text
* Simplify explicit exports without self-aliases
* fix: prevent repeating instructions in continued Responses API conversations
- Instructions are now only prepended to messages on the first turn
- When conversation_id/response_id exists (continuation), instructions are skipped
- Covers OpenAI and Azure Responses API paths
- Adds regression tests for all continuation scenarios
Fixes#3498
* Apply lint fixes to continuation tests
* Consolidate responses continuation tests
copilot-swe-agent[bot]andlarohra
Evan MattsonandGitHub
Tao ChenandGitHub
L. Elaine DazzioandGitHub
