概要
This skill applies the rigorous Red-Green-Refactor cycle of Test-Driven Development to the process of authoring AI instructions and documentation. By establishing baseline failures through pressure scenarios and iteratively refining skill definitions, it ensures that new capabilities are both effective and token-efficient. It serves as a comprehensive framework for developers looking to build robust, discoverable, and high-performance skills for Claude Code and Codex environments, focusing on the principle that if you haven't seen an agent fail without a skill, you cannot verify the skill's effectiveness.