writing-skills

Test-driven development for skills involves running a baseline RED test to capture initial failures, writing a minimal skill to pass tests, and iteratively refining the SKILL.md file until all deployment scenarios pass.

The Red-Green-Refactor workflow for skill documentation captures baseline failures, creates a minimal skill version, and closes gaps iteratively to ensure the documentation accurately reflects actual agent behavior and deployment readiness.

Rationalization tracking documents explicit rationalizations and red flags during skill testing to prevent common shortcuts, ensuring the iterative refinement process maintains integrity from baseline testing through deployment verification.

Yes, baseline testing without the skill is required before deployment to observe agent failures, document gaps, and establish a foundation for iteratively refining SKILL.md, including YAML frontmatter and description checks.

Structuring SKILL.md for deployment readiness requires proper YAML frontmatter, clear description fields, and progressive disclosure guidelines that streamline loading for Claude during deployment verification checks.

Yes, the test-driven workflow applies across skill creation, editing, and deployment verification, allowing you to run baseline tests, document failures, and iteratively refine existing SKILL.md files until scenarios pass.