What problem does it solve?
This Skill provides a structured workflow for building new Claude skills: capturing user intent, drafting clear SKILL.md documentation, generating and running evaluations, and iterating based on results to improve performance and triggering accuracy.
Core Features & Use Cases
- Capture intent from user conversations and translate it into a kebab-case skill name and a descriptive SKILL.md frontmatter.
- Write the SKILL.md body with progressive disclosure, including a concise workflow, recommended scripts, and references to reusable assets.
- Run automated evaluation loops (trigger tests) against a diverse set of prompts to measure trigger accuracy, evaluation results, and iteration progress.
- Iterate description and skill content automatically or semi-manually based on evaluation feedback, then re-run tests until performance stabilizes.
- Use the packaging and reference tooling to bundle scripts/assets for distribution and reuse in other projects.
Quick Start
Interview the user to capture intent, draft the SKILL.md, design and run evals, and iterate based on results.