Python: Fix ag-ui regressions

[moonbox3]

Motivation and Context

While introducing the AGUIChatClient, the related code changes affected the Dojo examples that were previously published. This PR fixes those regressions and now allows for the use of all seven examples, along with client tools, which previously wasn't working properly. More unit tests were added now for a 94% coverage across the ag-ui code.

Description

Fixes regressions. Adds more tests

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.

[markwallace-microsoft]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/ag-ui/agent_framework_ag_ui
_events.py	258	7	97%	146, 272, 291, 322, 565–566, 597
_message_adapters.py	130	30	76%	57, 87, 95–96, 98, 102–105, 118, 200, 202–203, 205–209, 211, 213–214, 237, 239, 241–242, 244–246, 259, 261
_orchestrators.py	370	30	91%	137, 454, 491, 521–522, 554, 616–618, 626–627, 633–634, 638–639, 653–654, 694–696, 703, 706–707, 710–713, 715–717
TOTAL	13830	2010	85%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
1856	109 💤	0 ❌	0 🔥	37.019s ⏱️

[Copilot]

Pull Request Overview

This PR fixes regressions in the AG-UI code that affected Dojo examples and client tool functionality. The changes improve tool result handling, message sanitization, and orchestration logic to properly support human-in-the-loop approval flows and declaration-only frontend rendering tools.

Key Changes:

• Fixed message adapter logic to properly distinguish between approval responses and tool results
• Added message sanitization to filter orphaned tool results and inject synthetic results for confirm_changes
• Fixed client tool merging to avoid overriding agent-configured tools with duplicate names
• Added comprehensive unit tests achieving 94% coverage

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
test_orchestrators_coverage.py	New comprehensive test file covering orchestrator edge cases and sanitization logic
test_agent_wrapper_comprehensive.py	Updated test expectations for orphaned tool result handling
main.py	Added startup logging messages for the AG-UI examples server
ui_generator_agent.py	Converted @ai_function decorators to AIFunction declarations for frontend-rendered tools
_orchestrators.py	Major refactor adding message sanitization, deduplication, and improved tool merging logic
_message_adapters.py	Refactored tool result message handling to fix approval response detection
_events.py	Added tracking for tool calls that have ended to prevent duplicate events
launch.json	Fixed module name for debugging configuration

Comments suppressed due to low confidence (1)

python/packages/ag-ui/tests/test_orchestrators_coverage.py:1

• This if False: block is used to make the async generator syntactically valid without yielding. A cleaner approach would be to use return or raise AssertionError(...) directly without the dead code block.

# Copyright (c) Microsoft. All rights reserved.