skill-creator

You can automate AI skill creation by drafting SKILL.md files from user prompts, running evaluation loops, and benchmarking to iteratively improve trigger accuracy. This workflow guides interviews and coordinates eval workflows to optimize skill development.

An evaluation loop is a benchmarking workflow that tests, evaluates, and iterates AI skills to improve trigger accuracy across prompts. It generates clear evaluation reports to measure performance and refine SKILL.md content.

You can convert ideas into structured SKILL.md frontmatter and content by using guided interviews. The skill creation process drafts the file structure and coordinates content generation based on user prompts.

Yes, the skill creation workflow requires YAML as a dependency to structure and manage the SKILL.md frontmatter. This ensures proper formatting and parsing of the skill metadata during the evaluation loops.

Yes, you can sync local skills with upstream repositories and re-synchronize AI skills when upstream changes occur. This ensures local skill versions remain consistent with the latest upstream updates.

The best way to benchmark AI skills is running end-to-end evaluation loops that test triggering across prompts, manage multi-iteration history, and generate benchmark reports to track accuracy improvements.