writing-skills

To write a discoverable SKILL.md file, draft a YAML frontmatter with a name, CSO-friendly description, and clear keywords. This standardizes the structure for Claude-like agents and optimizes it for search and recall.

Progressive disclosure in skill documentation is the practice of keeping the SKILL.md body concise while linking to optional scripts, references, and assets. Use it when you need to provide detailed guidance without exceeding token-safe metadata limits.

The best way to structure YAML frontmatter for agent skill files is to include a name and a concise description, supported by clear keywords. This enforces token-safe metadata limits and ensures the skill is testable.

Writing skill documentation does not require external dependencies. The skill unit standardizes SKILL.md frontmatter and guidance, allowing you to optionally link scripts and assets only when progressive disclosure is needed.

To make your Claude skill documentation easy to find in search, enforce CSO-friendly descriptions and clear keywords in the YAML frontmatter. This approach optimizes the SKILL.md unit for searchability and recall.

Yes, skill metadata has token-safe limits. The skill enforces token-safe metadata limits to ensure efficient processing, requiring a standardized YAML frontmatter and a concise body for guidance.