skills.rest
CRAG666
CRAG666COMMUNITY
Explore all skills by this author →
GitHub

skill-creator

Create and refine Claude skills with SKILL.md evaluation and benchmarking.

46|4|Updated Apr 8, 2023
Search Tags:#benchmarking#claude skills#evals#description optimization#LLM workflow#skill triggering
One-click install
npx skills add https://github.com/CRAG666/dotfiles --skill skill-creator-crag666
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/CRAG666/dotfiles/tree/main/skills/.claude/skills/skill-creator
Command: npx skills add https://github.com/CRAG666/dotfiles --skill skill-creator-crag666

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve?

Skill-creator helps you build a Claude skill that reliably triggers and performs well by turning your goals into a draft, then running test prompts and iterating based on quantified and qualitative results.

Core Features & Use Cases

  • Skill drafting from intent: Converts your description of what you want into a complete SKILL.md with appropriate structure and triggering guidance.
  • Evaluation and benchmarking loop: Builds an eval set, runs trigger tests, grades outputs, aggregates pass-rate and timing, and surfaces patterns.
  • Description optimization: Improves the skill’s frontmatter description so Claude triggers the skill in the right contexts without over-triggering.
  • Workflow you can steer: Supports creating from scratch or improving an existing skill, depending on where you are in the process.

Quick Start

Use the skill-creator skill when you want to turn an idea like "make a skill that does X" into a working Claude skill, including test cases and an iterative evaluation loop.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I create a Claude skill that reliably triggers in the right contexts?▼

To create a Claude skill that reliably triggers, you need description optimization to refine the SKILL.md frontmatter, ensuring Claude invokes the skill accurately without over-triggering in unrelated contexts.

What is the best way to evaluate and benchmark LLM workflow performance?▼

Evaluating LLM workflow performance involves building an eval set, running trigger tests, grading outputs via grader artifacts, and aggregating pass-rate and timing results into a benchmark summary to surface patterns.

How do I iterate on an existing skill to improve its triggering accuracy?▼

Iterate on an existing skill by running an evaluator loop that tests trigger checks against graded expectations, using quantified and qualitative results to optimize the description under a length constraint.

Can I draft a complete SKILL.md file from a basic intent description?▼

Yes, skill drafting from intent converts your description of what you want into a complete SKILL.md with appropriate structure, test prompts, and triggering guidance for your new skill build.

Do I need prior test prompts to start benchmarking a new skill?▼

No, the skill creation workflow builds the eval set and generates test prompts for you, applying structured workflows and graded expectations to quantify performance even when starting from scratch.

Related Skills

Explore More Software Engineering Skills→
coreyjacoreyja
12

skill-writer

Draft SKILL.md frontmatter and structure guidance for Claude Code skills.

Community
Basic
ouakarouakar

prompt-optimizer

Automate end-to-end prompt optimization for failing skills using DSPy.

Community
Advanced
dakesandakesan
2

skill-creator

Create or update Claude Skills with YAML frontmatter and SKILL.md content.

Community
Basic
charly-vibescharly-vibes

extract-prompt

Extract reusable prompt templates from AI conversations by analyzing successful patterns.

Community
Intermediate