What problem does it solve?
Create, refine, and validate Claude Skills efficiently by guiding users through drafting SKILL.md, running evals, benchmarking, and iterating based on results to achieve reliable triggering and performance.
Core Features & Use Cases
- Structured SKILL.md authoring: guides exact frontmatter fields, description formatting, and body organization to ensure discoverability and consistency.
- End-to-end eval loop: orchestrates trigger evaluations, collects results, and produces benchmark data for progress tracking.
- Iterative improvement workflow: proposes changes to descriptions and tests based on results and maintains a history for reproducibility and audits.
- Governance and replication: keeps a transparent record of iterations so teams can reproduce improvements across projects.
Quick Start
Create SKILL.md for a new skill, add optional scripts/references/assets, and run the loop to start evaluations.