Applies Test-Driven Development principles to the creation and validation of agent skills and documentation.
This skill provides a rigorous framework for authoring Claude Code skills by treating documentation as production code and agent behavior as test results. By following a strict Red-Green-Refactor cycle—writing pressure scenarios, observing baseline failures, and then drafting targeted documentation—developers ensure that their custom skills actually improve agent performance and aren't just redundant text. It emphasizes Claude Search Optimization (CSO) to ensure skills are discoverable and context-efficient.
Key Features
01101 GitHub stars
02Token-efficient documentation patterns
03Claude Search Optimization (CSO) guidelines
04Red-Green-Refactor cycle for documentation
05Pressure scenario testing with subagents
06Standardized SKILL.md structure templates
Use Cases
01Authoring new domain-specific skills for Claude Code
02Optimizing existing skills for better discovery and lower context usage
03Verifying agent compliance through baseline failure testing