Python: Fix responses agent kernel args bug (#13056)

### Motivation and Context

The `arguments: KernelArguments` are not getting handled properly for
the Responses Agent invoke. Use the provided `arguments` instead of
trying to get them from the kwargs.

<!-- Thank you for your contribution to the semantic-kernel repo!
Please help reviewers and future users, providing the following
information:
  1. Why is this change required?
  2. What problem does it solve?
  3. What scenario does it contribute to?
  4. If it fixes an open issue, please link to the issue here.
-->

### Description

- Closes #13053
- Adds unit test coverage. 

<!-- Describe your changes, the overall approach, the underlying design.
These notes will help understanding how your code works. Thanks! -->

### Contribution Checklist

<!-- Before submitting this PR, please make sure: -->

- [X] The code builds clean without any errors or warnings
- [X] The PR follows the [SK Contribution
Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
and the [pre-submission formatting
script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts)
raises no violations
- [X] All unit tests pass, and I have added new tests where possible
- [X] I didn't break anyone 😄

Author: moonbox3

python/semantic_kernel/agents/open_ai/responses_agent_thread_actions.py

@@ -256,7 +256,7 @@ async def invoke(
                     kernel.invoke_function_call(
                         function_call=function_call,
                         chat_history=override_history,
-                        arguments=kwargs.get("arguments"),
+                        arguments=arguments,
                         execution_settings=None,
                         function_call_count=fc_count,
                         request_index=request_index,
@@ -561,7 +561,7 @@ async def invoke_stream(
                     kernel.invoke_function_call(
                         function_call=function_call,
                         chat_history=override_history,
-                        arguments=kwargs.get("arguments"),
+                        arguments=arguments,
                         is_streaming=True,
                         execution_settings=None,
                         function_call_count=fc_count,

python/tests/unit/agents/openai_responses/test_openai_responses_thread_actions.py

@@ -19,6 +19,7 @@
 from semantic_kernel.contents.streaming_chat_message_content import StreamingChatMessageContent
 from semantic_kernel.contents.streaming_text_content import StreamingTextContent
 from semantic_kernel.contents.utils.author_role import AuthorRole
+from semantic_kernel.functions import KernelArguments
 
 
 @pytest.fixture
@@ -67,7 +68,6 @@ def mock_thread():
     return thread
 
 
-@pytest.mark.asyncio
 async def test_invoke_no_function_calls(mock_agent, mock_response, mock_chat_history, mock_thread):
     async def mock_get_response(*args, **kwargs):
         return mock_response
@@ -89,7 +89,6 @@ async def mock_get_response(*args, **kwargs):
     assert final_msg.role == AuthorRole.ASSISTANT
 
 
-@pytest.mark.asyncio
 async def test_invoke_raises_on_failed_response(mock_agent, mock_chat_history, mock_thread):
     mock_failed_response = MagicMock(spec=Response)
     mock_failed_response.status = "failed"
@@ -115,7 +114,6 @@ async def mock_get_response(*args, **kwargs):
             pass
 
 
-@pytest.mark.asyncio
 async def test_invoke_reaches_maximum_attempts(mock_agent, mock_chat_history, mock_thread):
     call_counter = 0
 
@@ -173,7 +171,6 @@ async def mock_get_response(*args, **kwargs):
         assert messages is not None
 
 
-@pytest.mark.asyncio
 async def test_invoke_with_function_calls(mock_agent, mock_chat_history, mock_thread):
     initial_response = MagicMock(spec=Response)
     initial_response.status = "completed"
@@ -227,6 +224,138 @@ async def mock_get_response(*args, **kwargs):
         assert len(messages) == 3, f"Expected exactly 3 messages, got {len(messages)}"
 
 
+async def test_invoke_passes_kernel_arguments_to_kernel(mock_agent, mock_chat_history, mock_thread):
+    # Prepare a response that triggers a function call
+    initial_response = MagicMock(spec=Response)
+    initial_response.status = "completed"
+    initial_response.id = "fake-response-id"
+    initial_response.output = [
+        ResponseFunctionToolCall(
+            id="tool_call_id",
+            call_id="call_id",
+            name="test_function",
+            arguments='{"some_arg": 123}',
+            type="function_call",
+        )
+    ]
+    initial_response.error = None
+    initial_response.incomplete_details = None
+    initial_response.created_at = 123456
+    initial_response.usage = None
+    initial_response.role = "assistant"
+
+    final_response = MagicMock(spec=Response)
+    final_response.status = "completed"
+    final_response.id = "fake-final-response-id"
+    final_response.output = []
+    final_response.error = None
+    final_response.incomplete_details = None
+    final_response.created_at = 123456
+    final_response.usage = None
+    final_response.role = "assistant"
+
+    responses = [initial_response, final_response]
+
+    async def mock_invoke_fc(*args, **kwargs):
+        # Assert that KernelArguments were forwarded
+        assert isinstance(kwargs.get("arguments"), KernelArguments)
+        assert kwargs["arguments"].get("foo") == "bar"
+        return MagicMock(terminate=False)
+
+    mock_agent.kernel.invoke_function_call = MagicMock(side_effect=mock_invoke_fc)
+
+    async def mock_get_response(*args, **kwargs):
+        return responses.pop(0)
+
+    with patch.object(ResponsesAgentThreadActions, "_get_response", new=mock_get_response):
+        args = KernelArguments(foo="bar")
+        # Run invoke and ensure no assertion fails inside mock_invoke_fc
+        collected = []
+        async for _, msg in ResponsesAgentThreadActions.invoke(
+            agent=mock_agent,
+            chat_history=mock_chat_history,
+            thread=mock_thread,
+            store_enabled=True,
+            function_choice_behavior=MagicMock(maximum_auto_invoke_attempts=1),
+            arguments=args,
+        ):
+            collected.append(msg)
+        assert len(collected) >= 2
+
+
+async def test_invoke_stream_passes_kernel_arguments_to_kernel(mock_agent, mock_chat_history, mock_thread):
+    class MockStream(AsyncStream[ResponseStreamEvent]):
+        def __init__(self, events):
+            self._events = events
+
+        async def __aenter__(self):
+            return self
+
+        async def __aexit__(self, exc_type, exc_val, exc_tb):
+            pass
+
+        def __aiter__(self):
+            return self
+
+        async def __anext__(self):
+            if not self._events:
+                raise StopAsyncIteration
+            return self._events.pop(0)
+
+    # Event that includes a function call
+    mock_tool_call_event = ResponseOutputItemAddedEvent(
+        item=ResponseFunctionToolCall(
+            id="fake-tool-call-id",
+            call_id="fake-call-id",
+            name="test_function",
+            arguments='{"arg": 123}',
+            type="function_call",
+        ),
+        output_index=0,
+        type="response.output_item.added",
+        sequence_number=0,
+    )
+
+    mock_stream_event_end = ResponseOutputItemDoneEvent(
+        item=ResponseOutputMessage(
+            role="assistant",
+            status="completed",
+            id="fake-item-id",
+            content=[ResponseOutputText(text="Final message after tool call", type="output_text", annotations=[])],
+            type="message",
+        ),
+        output_index=0,
+        sequence_number=0,
+        type="response.output_item.done",
+    )
+
+    async def mock_get_response(*args, **kwargs):
+        return MockStream([mock_tool_call_event, mock_stream_event_end])
+
+    async def mock_invoke_function_call(*args, **kwargs):
+        assert isinstance(kwargs.get("arguments"), KernelArguments)
+        assert kwargs["arguments"].get("foo") == "bar"
+        return MagicMock(terminate=False)
+
+    mock_agent.kernel.invoke_function_call = MagicMock(side_effect=mock_invoke_function_call)
+
+    with patch.object(ResponsesAgentThreadActions, "_get_response", new=mock_get_response):
+        args = KernelArguments(foo="bar")
+        collected_stream_messages = []
+        async for _ in ResponsesAgentThreadActions.invoke_stream(
+            agent=mock_agent,
+            chat_history=mock_chat_history,
+            thread=mock_thread,
+            store_enabled=True,
+            function_choice_behavior=MagicMock(maximum_auto_invoke_attempts=1),
+            output_messages=collected_stream_messages,
+            arguments=args,
+        ):
+            pass
+        # If assertions passed in mock, arguments were forwarded
+        assert len(collected_stream_messages) >= 1
+
+
 async def test_invoke_stream_no_function_calls(mock_agent, mock_chat_history, mock_thread):
     class MockStream(AsyncStream[ResponseStreamEvent]):
         def __init__(self, events):
@@ -294,7 +423,6 @@ async def mock_get_response(*args, **kwargs):
         assert collected_stream_messages[0].role == AuthorRole.ASSISTANT
 
 
-@pytest.mark.asyncio
 async def test_invoke_stream_with_tool_calls(mock_agent, mock_chat_history, mock_thread):
     class MockStream(AsyncStream[ResponseStreamEvent]):
         def __init__(self, events):