skill-authoring-guide

To structure agent-skills, you must validate SKILL.md frontmatter and organize resources into scripts, references, and assets directories. This enforces a concise, testable unit format for Cursor.

The skill authoring validation workflow uses TDD principles to verify Skill integrity with templates and tooling. It initializes units via init_skill.py and checks structure compliance using quick_validate.py.

To validate SKILL.md frontmatter during development, run the quick_validate.py script. This tool checks your frontmatter fields and resource layouts to ensure your Skill unit meets structural requirements.

Yes, you can use init_skill.py to initialize new Skill units with the correct directory structure. It scaffolds the required SKILL.md file and resource folders to verify Skill integrity from the start.

The best way to organize agent-skills resources is following the recommended layouts for scripts, references, and assets. This structure ensures skills remain concise, testable, and reusable across workflows.

SKILL.md validation may fail if optional agents or openai.yaml files conflict with frontmatter requirements. Ensure your UI metadata and optional configurations align with the structural templates provided.