What problem does it solve?
Building reliable AI assistant skills requires more than just writing prompts—it demands systematic testing, quantitative benchmarking, and iterative refinement to ensure consistent triggering and high-quality outputs. This Skill provides the complete toolkit to transform a rough idea into a production-ready, evaluated skill.
Core Features & Use Cases
- Structured Skill Creation: Capture user intent, draft SKILL.md files with proper frontmatter, and organize resources using progressive disclosure principles.
- Evaluation & Benchmarking: Generate test cases, run parallel with-skill and without-skill comparisons, aggregate pass rates and timing metrics, and launch interactive review viewers for human feedback.
- Description Optimization: Iteratively improve skill descriptions using train/test splits to maximize triggering accuracy without overfitting to specific queries.
- Use Case: A developer building a data extraction skill can use this to draft the skill, run 10 realistic test prompts, compare outputs against a baseline, analyze variance, and package the final
.skill file for distribution.
Quick Start
Use the skill-creator skill to build a new skill that extracts invoice data from PDF files and run evaluations to verify it handles both fillable and scanned forms correctly.
