Python: fix(ag-ui, core): fix Anthropic client with ag-ui and mcp arg issue

[moonbox3]

Motivation and Context

When using the Anthropic client with AG-UI integration, requests fail with:

TypeError: AsyncMessages.create() got an unexpected keyword argument 'store'

The AG-UI orchestrator passes store=True when metadata is present, but Anthropic's API doesn't support this parameter.

The fix is to filter out unsupported options (store, logit_bias, seed, frequency_penalty, presence_penalty, conversation_id) in _prepare_options before passing to the Anthropic API.

These options were already documented as unsupported in AnthropicChatOptions via None type annotations, but weren't being filtered at runtime.

Description

• Fixes Python: [Bug]: AG-UI integration with Anthropic Foundry Agent throws error #3278

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
_orchestrators.py	421	46	89%	113, 122, 229, 497, 557–559, 566–567, 575, 577–579, 583, 592, 625, 627–629, 631–634, 636, 638–639, 641, 643–648, 756–758, 762–767, 818–820, 823
packages/anthropic/agent_framework_anthropic
_chat_client.py	343	137	60%	51, 360, 383, 385, 398, 420–423, 473–474, 483, 485–486, 491, 508–509, 551, 566, 570–571, 587, 596, 598, 602–603, 644–646, 648, 661–662, 669–671, 675–677, 681–684, 695, 697, 719, 729, 751–757, 764–765, 773–774, 782–785, 792–793, 799–800, 806–807, 813, 821–823, 827, 834–835, 841–842, 848–849, 855, 863–866, 873–874, 893, 900–901, 920, 942, 944, 953–954, 960, 982–983, 989–990, 999–1009, 1016–1022, 1029–1035, 1042–1051, 1058–1061
packages/core/agent_framework
_mcp.py	388	57	85%	112, 174, 183, 246, 254, 275, 365, 432, 467, 469, 473–474, 476–477, 480–481, 484–485, 531, 546, 564, 605, 711, 724–729, 751, 770, 773–775, 790–791, 797–799, 818, 827, 830–832, 847–848, 852–856, 873–877, 1017
_tools.py	758	74	90%	226, 272, 323, 325, 353, 523, 555–556, 658, 660, 680, 698, 712, 724, 729, 731, 738, 771, 827–829, 870, 895–904, 910–919, 955, 965, 1206, 1543, 1630–1634, 1759, 1761, 1827, 1919, 1925, 1967–1968, 1981–1982, 2025, 2109, 2147–2148, 2176–2178, 2216–2217, 2227, 2284–2285, 2292–2293
TOTAL	17452	2700	84%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
3187	213 💤	0 ❌	0 🔥	1m 8s ⏱️

[Copilot]

Pull request overview

This PR fixes a bug where the Anthropic client fails with TypeError: AsyncMessages.create() got an unexpected keyword argument 'store' when AG-UI integration passes unsupported options. The fix filters out six options (store, logit_bias, seed, frequency_penalty, presence_penalty, conversation_id) that are not supported by Anthropic's API but may be passed by OpenAI-compatible orchestrators.

Changes:

• Added UNSUPPORTED_OPTIONS constant to define options not supported by Anthropic API
• Updated _prepare_options method to filter out unsupported options before making API calls
• Added conversation_id to AnthropicChatOptions TypedDict's unsupported options list
• Added comprehensive test to verify unsupported options are filtered while supported options remain

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
python/packages/anthropic/agent_framework_anthropic/_chat_client.py	Added UNSUPPORTED_OPTIONS constant and filtering logic in _prepare_options; added conversation_id to unsupported options in TypedDict
python/packages/anthropic/tests/test_anthropic_client.py	Added comprehensive test verifying all six unsupported options are filtered out

[moonbox3]

I am working on a refactor of some AG-UI code... incoming soon.

[moonbox3]

Closing. Superseded by #3322.