creating-skills

To structure a reusable AI agent skill, create a root directory named after your skill containing a mandatory SKILL.md file with YAML frontmatter, and optionally add scripts, references, and assets folders to support its functionality.

A SKILL.md frontmatter file requires name and description fields as mandatory metadata, while optional fields include license, compatibility, and additional metadata to ensure proper skill documentation and reuse.

The open standard format for authoring agent skills works by enforcing a standardized directory structure and YAML frontmatter validation within SKILL.md, ensuring consistent naming, metadata compliance, and reusable skill lifecycle management across catalogs.

Yes, you can update existing agent skills using this blueprint by applying the open standard to validate their SKILL.md frontmatter and directory structure, ensuring they meet the consistency requirements for sharing and cataloging.

The best way to organize scripts and references when creating an agent skill is to place them in dedicated optional directories alongside your SKILL.md file, separating executable logic and supporting documentation from the skill's main metadata.

No specific dependencies are required to create a skill using the open standard; you only need to define a SKILL.md file with the mandatory name and description frontmatter fields within a properly named directory structure.