What is 'triggering accuracy'?

It refers to how reliably Claude identifies when to use a specific skill based on your request. Skill Creator includes tools to make these descriptions more effective.

How does Skill Creator evaluate performance?

It runs parallel test cases using subagents—one with the new skill and one without—then compares the outputs based on user-defined assertions and qualitative review.

What is a Claude Code Skill?

A Claude Code Skill is a specialized capability defined in a SKILL.md file that provides domain-specific guidance, patterns, and tools to enhance Claude's performance on specific tasks.

Do I need to write complex JSON for evaluations?

No, Skill Creator handles the creation of eval_metadata.json and assertion structures through a conversational interview process.

Can I use Skill Creator to optimize existing skills?

Yes, it can snapshot current skills, run them against new test cases, and help you rewrite instructions or trigger descriptions to improve accuracy.

Skill Creator

Name: Skill Creator
Author: ValorVie

byValorVie

•

개발자 도구

Automates the end-to-end process of creating, benchmarking, and optimizing custom Claude Code skills through structured iteration.

Skill Creator is a powerful meta-utility designed to streamline the development of custom Claude Code skills. It guides users through a comprehensive lifecycle: from initial intent capture and SKILL.md drafting to rigorous performance evaluation. By running automated side-by-side benchmarks—comparing 'with-skill' results against baseline runs—it provides both qualitative insights and quantitative metrics. The skill also includes specialized routines for optimizing trigger descriptions and refining jargon-appropriate communication, ensuring your custom AI capabilities are reliable, effective, and easy to maintain.

주요 기능

01Automated SKILL.md generation following architecture best practices

023 GitHub stars

03Guided interview workflow to capture intent and edge cases

04Trigger description optimization to improve activation accuracy

05Quantitative assertion testing for objective skill validation

06Side-by-side performance benchmarking with subagent execution

사용 사례

01Developing specialized skills for internal proprietary frameworks

02Iteratively improving the accuracy and reliability of existing AI workflows

03Standardizing AI coding assistant behaviors across a development team

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add valorvie/custom-skills skill-creator

For use in Claude.ai and ChatGPT

Download Skill