skill-creator

Create Claude Skills by planning reusable components, scaffolding directories with scripts/init_skill.py, writing YAML frontmatter in SKILL.md, following kebab-case naming, and organizing entry points (140–160 lines) with references (150–500 lines each). This structured approach ensures technical compliance and distribution readiness.

Progressive disclosure keeps Skill entry points lean and focused while documenting deeper knowledge in separate reference documents. This pattern helps users understand core functionality immediately without overwhelming them with advanced details, improving usability across complex domains.

Yes. Design reusable scripts, references, and assets as modular components during planning, then package them for distribution. This approach reduces duplication, accelerates skill creation and updates, and maintains consistency across your Skill ecosystem.

Claude Skills require YAML frontmatter in SKILL.md, kebab-case directory naming, entry point sizing of 140–160 lines, and references organized at 150–500 lines each. The quick_validate tool and Python support packaging validation and initialization workflows.

Initialize your Skill with scripts/init_skill.py, edit SKILL.md and resources, package and validate using quick_validate, then iterate based on results. This workflow ensures your Skill meets technical requirements before release.

Start by understanding the skill domain with concrete examples, then identify reusable scripts, references, and assets that serve multiple use cases. Map dependencies and entry point logic before scaffolding, reducing rework during implementation.