skill-creator

Create, test, and optimize AI skills with SKILL.md and evaluation scripts.

Updated Aug 27, 2026

Search Tags:#prompt engineering #optimization #ai development #skill creation #evaluation #llm skills

Ready to run

npx skills add https://github.com/baksha97/dotfiles --skill skill-creator-baksha97

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/baksha97/dotfiles/tree/main/meta/skills/skill-creator
Command: npx skills add https://github.com/baksha97/dotfiles --skill skill-creator-baksha97

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill streamlines the entire lifecycle of creating and improving AI skills, from initial concept to optimized performance, reducing the complexity and time required for development.

Core Features & Use Cases

Skill Creation: Guides users through defining intent, writing SKILL.md, and structuring skill resources.
Iterative Improvement: Facilitates testing, evaluation, and refinement based on user feedback and performance metrics.
Description Optimization: Enhances skill triggering accuracy through automated prompt testing and analysis.
Use Case: You have an idea for a new AI capability, like summarizing meeting transcripts. Use this Skill to draft the SKILL.md, create test cases, run evaluations, and optimize the description so Claude reliably uses your new skill.

Quick Start

Use the skill creator to build a new skill for summarizing meeting notes.

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 test AI skills from scratch?▼

To create and test AI skills, you define the skill intent, write a SKILL.md file with YAML frontmatter and Markdown instructions, bundle scripts, and run evaluations using run_eval.py to measure performance. This handles the complete skill creation lifecycle.

What is the best way to optimize LLM skill triggering accuracy?▼

The best way to optimize LLM skill triggering accuracy is using automated description optimization via run_loop.py and improve_description.py, which test queries and refine the skill description to ensure Claude reliably triggers the correct skill.

How does skill evaluation work during AI development?▼

Skill evaluation works by executing test queries against your AI skill using run_eval.py and viewing results with eval-viewer. This process assesses performance metrics, enables iterative refinement of instructions, and validates skill behavior based on test outcomes.

Do I need pytest and anthropic dependencies to build AI skills?▼

Yes, you need the anthropic and pytest dependencies to build AI skills effectively. The anthropic package interfaces with the LLM for skill execution, while pytest facilitates running evaluations and testing skill performance during iterative development.

Can I bundle scripts and references directly within my AI skill?▼

Yes, you can bundle scripts and references directly within your AI skill. The skill creation process supports structuring these assets alongside the SKILL.md file, enabling the skill to execute complex logic and access necessary reference materials.

Why is my AI skill not triggering correctly in response to user queries?▼

Your AI skill may not trigger correctly because the description lacks optimization. You can resolve this by running automated description optimization via run_loop.py and improve_description.py to analyze test queries and refine the SKILL.md description for accurate triggering.