v0.2.7 — Pip skill paths and full wheel bundles

Bug fixes

• Skill loader paths after pip install (#13)
  SkillLoader.load_skill("category/skill_name") no longer resolves only under site-packages/skills/. The loader now searches, in order:

  1. An existing path on disk (absolute or relative skill directory)
  2. Roots in SKILLWARE_SKILL_PATH (multiple paths separated by your OS path separator)
  3. A skills/ folder in the current working directory (and up to six parent directories)
  4. Bundled registry skills shipped with the package

  If nothing matches, the error lists the paths that were tried.

Packaging

• Full skill bundles on PyPI — Wheels now include manifest.yaml, instructions.md, card.json, and skill data files, not only .py modules.
• MANIFEST.in grafts the skills/ tree; [tool.setuptools.package-data] uses skills = ["**/*"] so new registry skills do not require per-skill pyproject.toml edits.
• Registry layout — Empty __init__.py files under skills/ category packages (and skill folders where needed) so setuptools packages the complete tree.

Documentation

• Usage guide: Finding skills on disk
• Contributor notes for PyPI packaging (CONTRIBUTING.md, skill template README)

Upgrade notes

• Clone / editable install: Behavior is unchanged for normal development from the repo root.
• Pip consumers: Place project skills in ./skills/<category>/<name>/, or set SKILLWARE_SKILL_PATH to your skills root. Registry IDs like compliance/tos_evaluator still work and now load complete manifests from the installed package.
• New registry skills: Add empty __init__.py in new category and skill directories (enforced in tests); do not add entries to pyproject.toml per skill.

Verification

pip install skillware==0.2.7
pytest tests/   # for contributors cloning the repo