Python: Support response_format in GitHubCopilotAgent - #9810

Hi Eduard van Valkenburg (@eavanvalkenburg) , please take a look when you get a chance

Eduard van Valkenburg (@eavanvalkenburg) - the pipeline failure is fixed. Package Checks (3.11) failed because pyright could not resolve the from_updates overload in the streaming finalizer: options.get("response_format") was partially unknown. It is now typed as Any (a14d0ce), and pyright, ruff and the github_copilot tests are clean locally.

Could you please retrigger the pipeline when you get a chance? Thanks!

Eduard van Valkenburg (@eavanvalkenburg) - I moved the alias parsing into core as you suggested (790ac8e), and fixed the typing failure in the new aliased-model tests (3d1e5c8). Could you please retrigger the pipeline when you get a chance?

If everything is green, would it be possible to get this into the next release? We are waiting on structured output for the GitHub Copilot agent. Thanks!

Dineshsuriya D (@droideronline) Hard-coding by_name=False in shared response parsing overrides models configured with validate_by_name=True or populate_by_name=True. A model with item_count: int = Field(alias="itemCount") previously accepted {"item_count": 3} through AgentResponse.value and ChatResponse.value; this head now raises ValidationError. Please preserve each model’s configured field-name validation while still supporting alias-keyed output, and add a name-keyed regression test.

Eduard van Valkenburg (@eavanvalkenburg) - fixed in 6cdf1e0. Core now validates with the model's own configuration first, so name-keyed output (validate_by_name or populate_by_name) is accepted as before. Only if that fails does it retry with aliases only (pydantic 2.11+) to support alias-keyed provider output, and it re-raises the original error if both fail. Added regression tests for name-keyed output and for the original error on invalid output.

Dineshsuriya D (@droideronline) One correctness issue remains in the alias fallback: defaulted aliased fields can silently lose the provider’s value. With validate_by_alias=False and validate_by_name=True, item_count: int = Field(default=0, alias="itemCount") parses {"itemCount": 3} as item_count=0 because initial validation succeeds and the alias retry never runs. I reproduced this through both streaming and non-streaming GitHubCopilotAgent paths. Please enable alias validation while preserving configured field-name validation, and add a defaulted-aliased-field regression for both run modes.

Eduard van Valkenburg (@eavanvalkenburg) - good catch, fixed in b2dc326. Core now validates with a single pass using by_alias=True and leaves by_name to the model's own configuration. Aliases are always accepted, so a defaulted aliased field keeps the provider's value instead of silently using its default, and models configured with validate_by_name or populate_by_name still accept name-keyed output. The retry logic is gone. Added defaulted-aliased-field regressions for the core response and for both streaming and non-streaming GitHubCopilotAgent runs.

Dineshsuriya D (@droideronline) Please add a note to the PR description that this change depends on the next agent-framework-core version bump. The current agent-framework-github-copilot dependency floor still permits core 1.22.0, which does not contain the shared parser fix used by this PR. No dependency change is required here; the next version bump will update the floor.

Eduard van Valkenburg (@eavanvalkenburg) - added a dependency note to the PR description: this change depends on the next agent-framework-core version bump, and no dependency change is needed here since the bump will raise the floor.

Eduard van Valkenburg (@eavanvalkenburg) - addressed the three streaming comments (request correlation, aborted idle, autopilot idle) and rebased onto the latest main, so the branch is now a single commit. Please take another look and retrigger the pipeline when you get a chance. Thanks!

Fixed the failing typing check (Test Typing Checks (3.11), which also tripped merge-gatekeeper). The two new streaming tests set aborted and mode on the idle event data, which ty and mypy reject on the SDK's data union type. They now assign through an Any-typed variable. Test-only change in 397de11; the Copilot tests, ruff and ty are clean locally.