What is a Claude Code Skill?

A specialized capability that provides domain-specific guidance, best practices, and implementation patterns to enhance Claude's behavior during specific tasks.

How are test cases handled in the skill creation process?

The skill helps you define 2-3 realistic prompts, saves them to an evals file, and spawns subagents to test the skill's output both with and without the new instructions.

What is 'undertriggering' and how does this skill fix it?

Undertriggering is when Claude fails to use a relevant skill. Skill Creator optimizes the 'description' section of your skill to be more proactive in triggering for relevant tasks.

How does Skill Creator improve existing skills?

It snapshots your current skill, runs comparative benchmarks against a modified version, and uses quantitative assertions to measure performance gains.

Do I need to be a programmer to use Skill Creator?

No, the skill is designed to adapt its communication style based on your technical level, providing explanations for jargon like JSON or assertions when needed.

Skill Creator & Optimizer

Name: Skill Creator & Optimizer
Author: guanyang

byguanyang

•

417

•

개발자 도구

Streamlines the lifecycle of custom Claude Code skills through automated drafting, iterative benchmarking, and triggering optimization.

Skill Creator is a meta-utility designed to empower users to build, refine, and validate modular capabilities for Claude Code. It guides users through the entire development process—from capturing intent and drafting SKILL.md files to running side-by-side evaluations against baselines. By integrating quantitative assertions and qualitative reviews, it ensures that new skills are highly effective and optimized for triggering accuracy, helping both developers and non-technical users transform complex workflows into reusable AI expertises.

주요 기능

01Automated SKILL.md generation with optimized triggering descriptions

02Quantitative evaluation framework with customizable assertions

03417 GitHub stars

04Side-by-side performance benchmarking against baseline models

05Triggering optimization to ensure skills activate at the right time

06Iterative improvement loops for refining skill logic and output

사용 사례

01Measuring the performance impact of skill modifications using variance analysis

02Converting repetitive manual coding workflows into reusable modular skills

03Refining skill descriptions to prevent 'undertriggering' in complex environments

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add guanyang/antigravity-skills skill-creator

For use in Claude.ai and ChatGPT

Download Skill