skill-creator

Automate creation, testing, and refinement of Claude skills with SKILL.md and evals.

Updated Apr 6, 2026
One-click install
npx skills add https://github.com/yibozhong/skills --skill skill-creator-yibozhong
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/yibozhong/skills/tree/main/skill-creator
Command: npx skills add https://github.com/yibozhong/skills --skill skill-creator-yibozhong

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyyaml, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

The Skill unit helps teams systematically create, test, and improve Claude skills by providing a repeatable workflow for drafting SKILL.md, running trigger evaluations, benchmarking results, and iterating based on feedback.

Core Features & Use Cases

  • Draft, organize, and document a skill in SKILL.md, including frontmatter and operational instructions.
  • Run automated trigger-evaluation loops across train/test splits, and record results in a structured format.
  • Generate and analyze benchmarks, then iterate the skill description and scripts to improve triggering accuracy and effectiveness.

Quick Start

Draft the SKILL.md, prepare a small eval set, and run the evaluation loop to start iterating on a skill.

Frequently Asked Questions about skill-creator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate testing and refinement of Claude skills?

Automating Claude skill refinement involves running trigger-evaluation loops across train/test splits, collecting metrics, and iterating the skill description. This workflow specifies when to trigger and how to improve triggering accuracy using scripts and benchmark results.

What is the best way to structure a SKILL.md file for prompt engineering?

Structuring a SKILL.md file requires drafting frontmatter and operational instructions to document the skill. This process organizes the prompt engineering logic and specifies when the skill should trigger within the Claude environment.

How does an evaluation loop work when creating Claude skills?

An evaluation loop works by running automated trigger tests across train/test splits and recording results in a structured format. It then analyzes benchmarks to iterate the skill description and scripts, generating a final optimized skill with improved triggering accuracy.

Do I need pyyaml to run skill evaluations and benchmarks?

Yes, pyyaml is required to run skill evaluations and benchmarks. This dependency supports the scripts that manage the SKILL.md file, run automated evals, and collect metrics for the iteration workflow.

Can I use this workflow to generate an optimized skill description from benchmark results?

Yes, you can generate an optimized skill description from benchmark results. The workflow analyzes the collected metrics from the evaluation loops and iterates the skill description to improve triggering accuracy and overall effectiveness.