skill-creator

To package agent skills, you structure scripts, references, and assets into self-contained units. This process ensures AI agents can seamlessly import and execute the defined capabilities without missing dependencies.

Progressive disclosure in agent skill design patterns involves structuring SKILL.md to expose essential guidance first, revealing deeper details only as needed. This prevents context overload and optimizes agent performance.

Yes, YAML frontmatter containing the name and description fields is required in SKILL.md. This metadata validates the skill structure and provides AI agents with the necessary context to trigger the correct workflow.

You initialize a new agent skill by running the init_skill.py template script. This generates the foundational directory structure, allowing you to then customize SKILL.md with specific metadata and operational guidance.

Yes, you can include optional resources like scripts and references when packaging skills. The structure supports describing dependencies and components, ensuring the self-contained unit functions correctly within the target environment.

The best way to organize references and assets is by grouping them within the self-contained skill directory. This encapsulates necessary components, ensuring the AI agent can access all required resources without external path dependencies.