Anthropic's Agent Skills ecosystem leans Python (official doc skills use Python; the skills spec excludes node_modules from packaged distributions). Using Python + PyYAML aligns with where the platform is going, keeps the dependency declared via requirements.txt (scannable by pip-audit/Dependabot), and eliminates the need to vendor or bundle an npm package into the repo. Changes: - Rewrite src/scripts/resolve-customization.js as src/scripts/resolve_customization.py with identical merge semantics. - Drop vendored yaml@2.8.2 (~135 files, 800KB) from src/scripts/vendor/. - Declare the dependency via src/scripts/requirements.txt (PyYAML>=6.0). - Update all 6 agent SKILL.md files to invoke `python3` instead of `node`. - Update docs/how-to/customize-bmad.md to reflect the Python runtime and PyYAML install requirement. - Remove src/scripts/vendor/** from eslint.config.mjs and .prettierignore. - Refresh the _installSharedScripts comment (no more yaml bundling). |
||
|---|---|---|
| .. | ||
| docs | ||
| installer | ||
| build-docs.mjs | ||
| fix-doc-links.js | ||
| format-workflow-md.js | ||
| javascript-conventions.md | ||
| migrate-custom-module-paths.js | ||
| platform-codes.yaml | ||
| skill-validator.md | ||
| validate-doc-links.js | ||
| validate-file-refs.js | ||
| validate-skills.js | ||
| validate-svg-changes.sh | ||