feat(ai-engineering): add multi-platform API Example iteration workflow

[alienzh]

Summary

• Add a repository-level AI engineering skill for multi-platform API Example iteration.
• Introduce one shared Contract, independent Android/iOS/macOS/Windows Implementation agents, and separate Verification agents.
• Add the coverage matrix, knowledge index, acceptance gate, role routing, repository profile, and reusable manifest templates.
• Add tools for backlog generation, requirement initialization, Codex role dispatch, repository-delta attribution, and acceptance-manifest validation.
• Support repository-specific SDK sources and staggered platform targets, such as Android 4.6.3 with iOS/macOS/Windows 4.6.2.
• Require real Windows-host evidence for Windows build acceptance; macOS does not simulate or cross-compile Windows.

Acceptance Boundary

• Repository acceptance covers API Example source, platform builds, SDK-version consistency, review, parity, and UX evidence.
• Jenkins packaging, QA validation, artifact URLs, distribution, and website publication remain external processes.
• The Lead still controls phase advancement, retries, and final cross-platform acceptance.

Validation

• PYTHONDONTWRITEBYTECODE=1 python3 -m unittest discover -s docs/ai-engineering/tools -p '*_test.py' -v
  • 87 tests passed.
• Validated a real staggered SDK configuration:
  • Android: 4.6.3
  • iOS: 4.6.2
  • macOS: 4.6.2
  • Windows: 4.6.2
• All repository SDK-version checks passed.
• Acceptance manifest validation passed.
• git diff --check passed.

Scope

• No Android, iOS, macOS, or Windows example source changes.
• No Jenkins or CI pipeline script changes.