skill-creator

Automates the creation and iterative improvement of Skill documentation and evaluations.

Updated Feb 27, 2026
One-click install
npx skills add https://github.com/Crispy13/copilot-directions --skill skill-creator-crispy13
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/Crispy13/copilot-directions/tree/main/.github/skills/skill-creator
Command: npx skills add https://github.com/Crispy13/copilot-directions --skill skill-creator-crispy13

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a structured workflow for creating, evaluating, and iterating Claude skills, including capturing intents, interviewing, writing SKILL.md, and running evals to improve performance.

Core Features & Use Cases

  • Capture Intent: determine what the skill should enable Claude to do and how triggers should occur.
  • Interview & Research: ask edge-case questions and gather context and references.
  • Write SKILL.md: generate a structured frontmatter (name, description) and a body with step-by-step instructions, references, and examples.
  • Run Evaluations: draft prompts, run evals, and generate reviews using eval-viewer and grader tooling.
  • Iterate & Improve: rewrite and re-evaluate based on feedback, progressively refining prompts, tests, and documentation.
  • Documentation scaffolding: may scaffold evals, references, and assets for extensibility via scripts.

Quick Start

Outline the intended skill, write the SKILL.md, create initial evals, and iterate on feedback until the skill is solid.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I create and iterate on Claude skills with automated evaluations?

To create and iterate on Claude skills, capture your intent, write a SKILL.md file, run evaluations, analyze results, and refine prompts to improve triggering accuracy. This structured loop ensures progressive enhancement of skill performance.

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

The best way to write a SKILL.md file is to generate structured frontmatter with name and description, followed by a body containing step-by-step instructions, references, and examples. Interviewing for edge cases ensures robust prompt engineering.

Can I run evaluations on Claude skills to improve triggering accuracy?

Yes, you can run evaluations on Claude skills by drafting prompts and generating reviews using eval-viewer and grader tooling. Analyzing these results allows you to iterate and progressively refine prompts for better triggering accuracy.

How do I capture intent and edge cases for a new automation workflow?

Capture intent and edge cases for an automation workflow by determining what the skill should enable Claude to do, then conducting interviews to ask edge-case questions and gather context, references, and required triggers.

Does skill-creator scaffold documentation and evaluation assets?

Yes, skill-creator scaffolds documentation and evaluation assets. It can scaffold evals, references, and assets for extensibility via scripts, providing a structured foundation for further skill development and testing.