What problem does it solve?
This Skill helps you create new Claude-compatible skills and improve existing ones by running structured trigger evaluations, collecting quantitative results, and iterating on the skill’s description until it triggers accurately.
Core Features & Use Cases
- Skill creation guidance: Turns a user’s desired workflow into a draft skill, including trigger context, expected outputs, and test prompts.
- Evaluation + iteration loop: Generates an eval set, runs trigger tests, benchmarks performance, and iterates on the description using held-out test queries to reduce overfitting.
- Performance analysis tooling: Aggregates grading results and can generate a review viewer so humans can inspect outputs alongside benchmark metrics.
- Description optimization: Improves the YAML frontmatter description to increase correct skill triggering.
- Use cases: Create a skill from scratch, modify an underperforming skill, validate that it triggers on the right user intents, and benchmark variance/consistency.
Quick Start
Use the skill creator to improve an existing skill by pasting the skill’s SKILL.md content and telling it what kinds of user prompts should trigger the skill.
