skill-creator

You can draft, test prompts, and iterate on skill logic end-to-end by creating a new SKILL.md with frontmatter name and description, then running the evaluation loop to capture metrics and compare iterations.

Optimizing skill descriptions for triggering accuracy involves using structured evaluation workflows to test, capture metrics, and compare iterations, ensuring the description boosts user understanding and performance.

Running evals improves CLI skill creation by capturing performance metrics across multiple iterations, allowing you to rapidly compare results and refine triggers within your automation workflow.

Yes, pyyaml is a required dependency to parse the YAML frontmatter in your SKILL.md file, enabling the structured evaluation workflows needed to test and iterate on skill logic.

Yes, you can refine triggers for existing skills by running the evaluation loop to capture metrics, compare new iterations against previous versions, and surface targeted improvements.