skill-creator

Automating skill creation involves drafting SKILL.md, configuring eval sets, and running trigger-evaluation loops to test benchmarks. This Skill orchestrates that end-to-end workflow, iterating on aggregated results until a high-quality, trigger-accurate skill is achieved.

An eval-driven development loop is a process that tests skill triggering and benchmarks against configured eval sets. It analyzes aggregated results and visual reports to iterate on skill descriptions and instructions until trigger accuracy is optimized.

Packaging finalized skills for distribution involves organizing the refined SKILL.md, accompanying scripts, references, and assets. This Skill automates the packaging workflow to facilitate adoption and deployment across multiple projects.

Yes, running end-to-end skill evaluation loops requires both the anthropic and pyyaml dependencies. These libraries support the underlying automation scripts and configuration parsing needed to benchmark and iterate on skill designs.

The best way to optimize skill descriptions for triggering is by running automated trigger-evaluation loops. This process benchmarks description variations, generates visual reports from aggregated results, and applies automated improvements to refine accuracy.