skill-creator

You create reusable AI skills by authoring a SKILL.md file with required YAML frontmatter and using progressive disclosure patterns. This approach separates core instructions from reference materials, preventing context bloat while extending Claude capabilities.

Progressive disclosure in skill development is the practice of keeping only essential metadata and high-level guidance in SKILL.md while moving detailed scripts and assets to separate reference files. Use it to prevent context bloat when authoring complex AI skills.

You validate and package a SKILL.md file for distribution by running the provided helper scripts. These scripts check for required YAML frontmatter fields, validate resource organization, and bundle the validated components into a distributable .skill package.

Yes, you need pyyaml installed because the validation helper scripts rely on it to parse and verify the mandatory YAML frontmatter fields in your SKILL.md file. This dependency ensures your metadata meets packaging requirements.

The best way to initialize a new skill project structure is to run the init_skill.py helper script with a kebab-case skill name and target path. This generates a template SKILL.md and example resource files to start authoring immediately.

Include scripts and references in your SKILL.md packaging when your skill requires automation logic or supplementary documentation. Keep core instructions in SKILL.md and move detailed processing scripts and reference data to separate resource files to maintain context efficiency.