About
This skill adapts the rigorous Red-Green-Refactor cycle to the creation of Claude Code skills, ensuring that documentation effectively guides AI agents in real-world scenarios. By requiring authors to observe an agent failing a task before writing the skill, it eliminates guesswork and ensures that instructions directly address specific failure modes and logic gaps. It provides a standardized framework for building robust, token-efficient, and highly discoverable reference guides that enhance Claude's capabilities across any project.