[bokelley]

Typed LegacyFormat inputs lost their VAST/DAAST versions during structural projection, even though the equivalent dictionary projected correctly. Normalize Pydantic models before collecting versions so nested typed requirements and root-model version lists follow the same registry resolution and advisory behavior as dictionary inputs.

Refs #1750. Generated models, schema pins and registry precedence are unchanged. Missing versions still fail closed.

Give the context-ID integration test fixture the unique private name _ContextIdEchoHandler. This prevents another test module’s _EchoHandler registration from hiding create_media_buy during full-suite collection. The test assertions and library behavior are unchanged.

Validation

• Full local gates in Docker: make lint, make typecheck-all, and make test passed. 18595 passed, 2729 skipped, 10 deselected, 1 xfailed, 1017 warnings in 3745.97s (1:02:25); coverage 92.14%.
• Docker: 96 tests passed across v1 projection, round trips and the registry.
• The five new versioned-model parity cases failed before the fix; both missing-version controls passed.
• Normal commit hooks passed: formatting, Ruff, source/adopter type checks, Bandit and file checks.
• Independent code and security reviews found no remaining issues.

SDK 10 forward port: #1759.

[aao-secretariat]

Ladon verdict: Approve

Approve — clean, well-scoped bug fix with no blocking or medium findings.

What I checked:

• The fix normalizes BaseModel inputs via model_dump(exclude_none=True) in _v1_version_constraints before walking, so typed LegacyFormat inputs no longer lose VAST/DAAST versions during structural projection (v1_to_v2.py:180). This restores the function's documented "tolerates both dict and model" contract.
• Parity with the dict path is asserted directly (from_model == from_dict), and fail-closed behavior on versionless formats is preserved and tested (test_typed_versionless_format_still_fails_closed).
• Private helper only — no public-surface, wire-shape, or discriminated-union change. The fix: conventional-commit prefix is correct; no semver break involved.
• No type-system import layering, generated-code, credential, or CI-gate concerns touched. New test coverage added for the fixed branch.

Decision table: no critical/high/medium findings; gated_paths is false; high_risk is false; no no-auto-approve team match; no prior decision. Falls through to row 9.

Note: review_decision is REVIEW_REQUIRED, but since gated_paths is false, row 2 does not apply and this does not block approval.

[aao-secretariat]

Ladon verdict: Approve

Approve — behavior-neutral test rename, no blocking or medium findings.

This is a subsequent review on a prior approve. The delta since the prior pass is a single commit renaming the test-local handler _EchoHandler → _ContextIdEchoHandler in tests/integration/test_a2a_context_id.py to prevent another module's _EchoHandler registration from shadowing create_media_buy during full-suite collection. The reviewer confirms the rename is complete (all 5 sites: 1 def + 4 call sites renamed, no stray references; private, not exported) and behavior-neutral.

No new findings. No critical/high/medium findings surfaced. Changes are confined to canonical-format translation code plus tests — no public-API/export changes, no generated-code edits, no gated paths, no high-risk files.

Decision table: rows 1–8 do not fire (no findings, gated_paths false, high_risk false, no team gate, prior decision was approve not escalate). Falls through to row 9 → approve.