skill-creator

Automating skill creation involves drafting a base skill, running structured evaluations, and iterating descriptions based on benchmarks to improve triggering accuracy. This provides a standardized frontmatter and body structure for repeatable workflows.

Standardizing frontmatter and body structure requires applying a repeatable workflow with optional bundles like scripts, references, and assets. This ensures consistency and robustness across rapidly prototyped skills through safety validations.

Running structured evaluations involves collecting qualitative feedback and generating quantitative benchmarks. You use these results to automatically improve skill descriptions, increasing triggering reliability and overall performance.

Yes, you can use PyYAML and Anthropic dependencies to draft, test, and benchmark skills. The system supports optional bundles including scripts, references, and assets while performing safety validations during the iteration process.

Triggering accuracy degrades when skill descriptions lack optimization based on evaluation results. Iteratively refining descriptions using quantitative benchmarks and qualitative feedback ensures stable improvements and reliable triggering performance.

Yes, markdown formatting is required to define the standardized frontmatter and body structure for new skills. This format enables the automated system to draft, test, and benchmark skills effectively throughout the iteration loop.