skill-creator

To create and package Claude skills, draft a SKILL.md file with YAML frontmatter and use a skill-creator workflow to validate, benchmark, and bundle it into a portable .skill file for deployment.

Evaluating skill trigger performance requires running trigger evaluations and benchmarks using an integrated run_eval scaffolding workflow to quantify when a skill should activate and measure its response accuracy.

Yes, PyYAML is a required dependency for parsing the YAML frontmatter in SKILL.md files, which defines the skill metadata and trigger conditions needed for the automated evaluation and packaging workflow.

Structure a skill folder by placing a SKILL.md file at the root alongside optional scripts/, references/, and assets/ directories to organize resources before running the draft, test, and packaging workflow.

Yes, you can iteratively improve existing skill descriptions and generate stronger triggers using the integrated eval and benchmark workflow, which validates changes before repackaging the updated .skill file.

Automated skill benchmarking workflows rely on a defined eval schema and structured YAML frontmatter in SKILL.md, meaning unstructured documentation or missing trigger definitions will cause validation and benchmark measurement failures.