What problem does it solve?
Building effective AI skills is hit-or-miss without systematic testing. Skills may fail to trigger, produce inconsistent outputs, or have descriptions that don't match real user queries. This skill provides the complete workflow for creating, evaluating, and refining skills with quantitative rigor.
Core Features & Use Cases
- End-to-End Skill Creation: Guide users from intent capture through draft, testing, and iteration for both new and existing skills.
- Quantitative Benchmarking: Run controlled experiments comparing with-skill vs without-skill performance, tracking pass rates, timing, and token usage with statistical variance analysis.
- Description Optimization: Automatically tune skill descriptions using train/test eval loops to improve triggering accuracy across diverse user queries.
- Use Case: A developer building a code review skill can capture their review workflow, draft the skill instructions, generate test cases from real PRs, run blind comparisons against baseline behavior, and optimize the description so Claude reliably invokes it when users ask for reviews.
Quick Start
Use the skill-creator skill to build a new skill for automating code review workflows by capturing your review process, drafting the SKILL.md instructions, and running benchmark tests to measure its triggering accuracy and output quality.
