Applies Test-Driven Development principles to the creation, verification, and optimization of Claude Code skills.
This skill introduces a rigorous Red-Green-Refactor framework for authoring Claude Code skills, treating documentation as production code. It ensures that every skill added to an agent's repertoire is necessary and effective by requiring developers to first observe an agent fail a task (Red), write the skill to rectify the behavior (Green), and then optimize the documentation for token efficiency and searchability (Refactor). This methodology eliminates 'placebo' documentation and ensures skills are triggered only under the correct conditions using Claude Search Optimization (CSO) patterns.
Key Features
01Red-Green-Refactor workflow for skill authoring
02Token-efficient progressive disclosure patterns
03Pressure scenario testing using subagents
04Claude Search Optimization (CSO) for better discovery
05Strict 'When-to-Use' description validation
061 GitHub stars
Use Cases
01Codifying complex team-specific coding standards for AI
02Creating reusable domain-specific techniques and patterns
03Reducing AI hallucinations through documented constraints
