skill-creator

To create an AI skill, follow a structured process from understanding its purpose to iterating on the design. You need to define the skill's anatomy, write the SKILL.md, and organize bundled resources like scripts and references.

Progressive disclosure is a core principle for AI skill creation that ensures skills remain concise and efficient. It involves setting degrees of freedom and structuring documentation so agents receive information contextually rather than all at once.

Structure AI agent documentation using a specific anatomy: a main SKILL.md file accompanied by organized directories for scripts, references, and assets. This structure ensures the agent can efficiently process workflow designs and capabilities.

You need to bundle scripts and references when your AI skill requires external documentation or automated actions to interact with specific APIs. Proper resource management organizes these assets to expand agent capabilities without cluttering the main instructions.

The best way to update existing AI skills is to iterate on their design by re-evaluating core principles like conciseness and progressive disclosure. This ensures the workflow design remains effective and well-documented for agent capabilities.

Yes, you can use this skill creation process to build capabilities for internal API interactions. It guides you in structuring the skill, writing instructions, and organizing necessary API documentation within the references directory.