.Net: Python: Bug: Expected message to be of type 'StreamingChatMessageContent', but got 'ChatMessageContent' instead

[smokhasi]

Describe the bug
When using input_transform/agent_callback with the latest version of semantic kernel in python,
running into this error.

• Tried variations of using streaming_agent_response_callback instead of agent_response_callback and run into the same error.
• Tried variations of returning a StreamingChatMessageContent from input_transform function but that runs into the same error.

2025-06-25 22:37:19 [ERROR] root: Error during orchestration: Expected message to be of type 'StreamingChatMessageContent', but got 'ChatMessageContent' instead.

To Reproduce

Sample script to reproduce the issue

class AgentInput(BaseModel):
    """
    Model class for the input parameters required by the agent.
    """
    run_id: str = Field(..., description="The unique identifier for the run.")
    target_id: str = Field(..., description="The unique identifier for the target.")
    branch_name: str = Field(..., description="The name of the github branch")


def custom_input_transform(input_message: AgentInput) -> ChatMessageContent:
    """
    Custom input transformation function that converts the AgentInput model into a ChatMessageContent object.
    """
    return ChatMessageContent(
        role=AuthorRole.USER,
        content=input_message.model_dump_json(),
    )


def agent_response_callback(message: ChatMessageContent) -> None:
    print(f"Agent {message.name}-{message.role} Response: '{message.content}'")


    sequential_orchestration = SequentialOrchestration[AgentInput, ChatMessageContent](
        members=agents,
        input_transform=custom_input_transform,
        agent_response_callback=custom_agent_response_callback
    )

    
    try:
        runtime = InProcessRuntime()
        runtime.start()

        orchestration_result = await sequential_orchestration.invoke(
            task=agent_input,
            runtime=runtime,
        )
        value = await orchestration_result.get(timeout=20)
        logging.info(f"\n------\n***** Final Result *****\n{value}")

        return response

    except Exception as e:
        logging.error(f"Error during orchestration: {e}")
    finally:
        # 5. Stop the runtime when idle
        await runtime.stop_when_idle()

Expected behavior
Should accept either ChatMessageContent or StreamingChatMessageContent object.

Platform

• Language: python
• Source: 1.34.0
• AI model: Azure OpenAI-4.1
• API version: 2025-04-01-preview
• using responses api
• IDE: VS Code
• OS: Mac

Additional context
Worked fine on 1.32.2

[moonbox3]

I can take this over, @eavanvalkenburg.