Python: fix: always include output in function call result message

[Hedrekao]

Motivation and Context

Previously the "output" field was only appended to function call result message in AzureOpenAIResponsesClient only if it was a 'truthy' value.

This meant that if the result of the tool call was an empty list (treated as 'falsy' value by python), the message was missing an output field, which was resulting in a following BadRequest error coming from OpenAI api:
ServiceResponseException: <class 'agent_framework.azure._responses_client.AzureOpenAIResponsesClient'> service failed to complete the prompt: Error code: 400 - {'error': {'message': "Missing required parameter: 'input[3].output'.", 'type': 'invalid_request_error', 'param': 'input[3].output', 'code': 'missing_required_parameter'}}.

Screenshot below showcases the error (I added a simple log to prints content.result.

The error comes from the fact that to according OpenAI Responses api schema, the "output" field is required as part of Function call tool output input message (link to schema)

The error can be simply reproduced by using a dummy tool that always returns empty list or None.

Description

This pull request simply always appends an "output" field to the function call output message, regardless whether the result of function call is truthy or not. I also made the mock tool and tested that the serialization function prepare_function_call_results is able to deal with None value as well as other falsy values.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the Contribution Guidelines
• All unit tests pass, and I have added new tests where possible
• Is this a breaking change? If yes, add "[BREAKING]" prefix to the title of the PR.

[Copilot]

Pull request overview

This PR fixes a bug where the "output" field was conditionally added to function call result messages only when the result value was truthy. This caused API errors when function calls returned falsy values (like empty lists, None, False, etc.), as the OpenAI Responses API requires the "output" field to always be present.

• Changed the conditional logic to always include the "output" field in function call result messages
• The fix applies to both OpenAIResponsesClient and AzureOpenAIResponsesClient (which inherits from the base class)
• The prepare_function_call_results function properly handles all falsy values by JSON-serializing them

[markwallace-microsoft]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/core/agent_framework/openai
_responses_client.py	412	70	83%	144–145, 148–149, 155–156, 159, 166, 201, 231, 259–260, 287, 291, 308, 313, 355, 419, 494, 499, 503–505, 525, 540–541, 545–547, 595, 615–616, 629–630, 646–647, 685, 687, 725, 727, 736–737, 750, 752, 825–831, 848–853, 872, 890, 900, 902, 920–921, 923–925, 936–937, 940, 942
TOTAL	15459	2352	84%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
2221	130 💤	0 ❌	0 🔥	55.576s ⏱️