Failing scenario using function tools with human in the loop approvals

[jcatlantis]

I’m trying to reproduce the “Using function tools with human-in-the-loop approvals” example from the documentation:
https://learn.microsoft.com/en-us/agent-framework/tutorials/agents/function-tools-approvals?pivots=programming-language-python

I’ve found that when using AzureAIAgentClient, the agent’s response text after approval is empty. However, when using AzureOpenAIChatClient, the result is as expected.

Does anyone know why this happens?

I’m using the latest version of the agent_framework Python package (1.0.0b251114).
Here is the code to reproduce the issue:

`

import asyncio
import os
from typing import Annotated

from agent_framework import (
    ChatMessage,
    Role,
    ai_function,
)
from agent_framework.azure import AzureAIAgentClient
from azure.identity.aio import AzureCliCredential


@ai_function(approval_mode="always_require")
def get_weather_detail(location: Annotated[str, "The city and state, e.g. San Francisco, CA"]) -> str:
    """Get detailed weather information for a given location."""
    return f"The weather in {location} is cloudy with a high of 15°C, humidity 88%."


async def main():
    agent = AzureAIAgentClient(
        async_credential=AzureCliCredential(),
        project_endpoint=os.environ.get("AZURE_AI_PROJECT_ENDPOINT"),
        model_deployment_name=os.environ.get("AZURE_AI_MODEL_DEPLOYMENT_NAME"),
    ).create_agent(
        name="WeatherAgent",
        instructions="You are a helpful weather assistant.",
        tools=[
            get_weather_detail,
        ],
    )

    response = await agent.run("What is the detailed weather like in Amsterdam?")
    if response.user_input_requests:
        for user_input_needed in response.user_input_requests:
            print(f"Function: {user_input_needed.function_call.name}")
            print(f"Arguments: {user_input_needed.function_call.arguments}")

            user_approval = True

            approval_message = ChatMessage(
                role=Role.USER,
                contents=[user_input_needed.create_response(user_approval)],
            )

            final_result = await agent.run(
                [
                    "What is the detailed weather like in Amsterdam?",
                    ChatMessage(role=Role.ASSISTANT, contents=[user_input_needed]),
                    approval_message,
                ]
            )

            print(final_result.text)


if __name__ == "__main__":
    asyncio.run(main())

`

[moonbox3]

@dmytrostruk or @eavanvalkenburg do you have an idea?

[eavanvalkenburg]

my initial thought, is that because of the thread in AzureAIAgents, you wouldn't have to reinput the user message and the assistants message, but just send the approval, which is then replaced with the function result, can you try with a similar setup as this one: https://github.com/microsoft/agent-framework/blob/main/python/samples/getting_started/tools/ai_function_with_approval_and_threads.py @jcatlantis

[jcatlantis]

Thanks for the reply.
And yes, using threads it works as expected.
However i dont get why should i be forced to explicitly use one.
Anyhow, it is working and i assume some fixes will come soonish ;)