creating-agent-skills

To structure a SKILL.md file for Claude agent skills, place it at the root of your skill directory and include YAML frontmatter with name and description fields. This skill provides centralized guidelines for frontmatter, XML zoning, and overall structure.

Progressive disclosure in skill design is a pattern that safely reveals complexity only when needed. You should use it when structuring robust Claude skills to ensure clear structure, reuse references, and prevent model overload.

To build a router pattern for AI workflows, follow the core principles and templates provided by this skill. It guides you through designing router patterns that help Claude efficiently navigate and select appropriate skill references.

The best way to audit and upgrade existing Claude skills is to use practical workflows that check for clear structure, proper XML zoning, and safe progressive disclosure. This skill provides checklists to validate and test model compatibility.

Yes, you need XML zoning references to create valid agent skills because they ensure clear structure and safe boundaries within the SKILL.md file. This skill provides templates for handling these references correctly.

Your Claude skill fails validation without YAML frontmatter because the system identifies valid units by verifying the presence of name and description fields in the YAML. The skill must have a SKILL.md at its root to be recognized.