skill-creator

To structure an AI skill, you organize a core SKILL.md file alongside bundled resources like scripts, references, and assets to deliver modular, specialized procedural knowledge and tool integrations for AI agents.

Progressive disclosure in AI skill creation is a core design principle for efficient context management, organizing content so agents receive necessary information layer by layer without overwhelming the operational context window.

The best way to write a SKILL.md file is to prioritize conciseness and appropriate degrees of freedom, structuring procedural knowledge and domain expertise so the AI agent can execute specialized tasks predictably.

You integrate an external API by creating a new skill, writing the procedural logic in the SKILL.md, placing executable interactions in the scripts directory, and organizing necessary API documentation within the references directory.

Skill creation requires understanding prompt engineering concepts to define appropriate degrees of freedom, but this developer guide provides a step-by-step framework from initial design to iteration for building effective AI extensions.

You should avoid modular skill structures when a task requires minimal procedural knowledge or lacks the need for domain expertise, as over-structuring simple prompts can introduce unnecessary context management overhead.