feat(validation): add frontmatter schema validator and standalone test suite

[poorvith-mp]

Summary

• Added �alidate_frontmatter_dict in skillware/version_policy.py to validate required metadata fields (
  ame, description).
• Added standalone unit test suite in ests/test_version_and_manifest.py.

Closes #320, closes #321, closes #322.

[rosspeili]

Thanks for wanting to help @poorvith-mp, I really appreciate the contribution.

I am going to close PR #323 and issues #320–#322 without merging. The work doesn’t match how Skillware is built today, and the PR doesn’t actually implement what those issues describe.

Why we’re closing the issues

• fix(cli): graceful fallback when optional pyyaml dependency is absent during manifest parsing #320 (PyYAML fallback): PyYAML is a core dependency (pyproject.toml), not optional. The CLI and loader import it at startup; manifest parsing uses full YAML in manifest.yaml, not a simple line parser. There’s no supported “PyYAML absent” install path to fix.
• feat(validation): add schema and frontmatter validation checks for skill instructions #321 (instructions frontmatter): Bundled skills don’t use YAML frontmatter in instructions.md, metadata lives in manifest.yaml, and instructions are markdown cognition per CONTRIBUTING. Validation for manifests, issuers, and non-empty instructions already exists via tests/test_skill_issuer.py, tests/test_registry_identity.py, tests/test_loader.py, and wheel smoke.
• feat(testing): add standalone unittest suite for version policy and manifest loading #322 (version/manifest tests): Version policy is already covered in tests/test_version_policy.py (pytest, including the CLI advisory hook). Manifest loading is covered in tests/test_loader.py and related framework tests. A separate unittest file would duplicate that without adding manifest coverage.

Why we can’t accept the PR

The diff adds validate_frontmatter_dict() in version_policy.py but nothing calls it, no loader, CLI, or CI integration, so it doesn’t change runtime behavior. It also doesn’t touch cli.py (#320), doesn’t match our instructions model (#321), and the new tests overlap existing pytest coverage without testing manifest loading (#322). Framework changes here would need wiring, pytest-style tests in the existing files, and a CHANGELOG entry if user-visible.

If you’d like to contribute in this area, a better fit would be a single, focused issue, for example extending manifest or instructions checks in the loader/issuer tests — after reading CONTRIBUTING.md and docs/contributing/ai_native_workflow.md. Happy to review something scoped that way.

[poorvith-mp]

Thanks for the clear architectural breakdown @rosspeili — makes complete sense regarding manifest.yaml vs instructions separation, PyYAML being a core dependency, and the existing pytest coverage. I'll read through docs/contributing/ai_native_workflow.md and keep any future PRs tightly scoped to existing test patterns.