skills.rest
garochee33
garochee33COMMUNITY
Explore all skills by this author →
WebsiteGitHub

skill-creator

Draft SKILL.md content, generate eval sets, and run benchmark loops.

3|Updated Apr 18, 2026
Search Tags:#benchmarking#skills#claude#evaluation#packaging#iteration
One-click install
npx skills add https://github.com/garochee33/DSH --skill skill-creator-garochee33
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/garochee33/DSH/tree/main/kb/claude/skills/skill-creator
Command: npx skills add https://github.com/garochee33/DSH --skill skill-creator-garochee33

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill helps you create new Claude skills from scratch, refine existing skills, and measure whether they actually work, so you can iterate on a skill with evidence instead of guesswork.

Core Features & Use Cases

  • Skill drafting: Turn a rough idea into a well-structured SKILL.md with clear trigger guidance and operational instructions.
  • Evaluation design: Create realistic test prompts, run comparisons, and review qualitative outputs alongside quantitative results.
  • Iteration and optimization: Use benchmark feedback to improve the skill content, reduce failures, and tighten the trigger description for better discovery.
  • Packaging: Bundle a finished skill into a distributable .skill file for installation and reuse.

Quick Start

Use the skill-creator skill to help me write a new skill for [task] and then test and refine it until it performs reliably.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I build and test Claude skills locally before packaging?▼

To build Claude skills, you draft a SKILL.md file, generate evaluation sets, run benchmark loops, and package the output into a .skill file using local Python scripts.

What is the best way to evaluate Claude skills and measure reliability?▼

Evaluating Claude skills involves creating realistic test prompts, running benchmark comparisons, and reviewing both qualitative outputs and quantitative results to measure actual reliability.

How do I improve a Claude skill that fails to trigger or performs unreliably?▼

Improving a Claude skill requires running benchmark loops to gather feedback, refining the SKILL.md content to reduce failures, and tightening trigger descriptions for better discovery.

Do I need Python installed to create and benchmark Claude skills?▼

Yes, creating and benchmarking Claude skills requires a local Python environment with the pyyaml dependency to run validation, benchmarking, reporting, and packaging scripts.

Can I turn a rough idea into a structured SKILL.md file automatically?▼

You can draft a structured SKILL.md from a rough idea by defining clear trigger guidance and operational instructions, then iterating on the content using benchmark feedback.

How does skill packaging work for redistributable Claude skills?▼

Skill packaging bundles a finished Claude skill into a distributable .skill file, allowing users to install and reuse the skill across different environments.

Related Skills

Explore More Software Engineering Skills→
mamd69mamd69
4

skill-creator

Automates drafting, testing, benchmarking, and iterating Claude skill SKILL.md workflows with Python and anthropic/pyyaml libraries.

Community
Advanced
minhtungominhtungo

skill-creator

Create standardized Claude Skill Units with YAML frontmatter and packaging.

Community
Intermediate
cuoreinpacecuoreinpace

llm-evaluation

Evaluate LLM applications with automated metrics, human feedback, and benchmarking.

Community
Advanced
NGTPTENGTPTE

skill-creator

Create, test, and improve reusable Claude Code skills from a draft SKILL.md.

Community
Advanced