skill-creator

Create and tune Claude skills through iterative evaluation and benchmarking.

Updated Jun 8, 2025
One-click install
npx skills add https://github.com/noartem/kawa --skill skill-creator-noartem
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/noartem/kawa/tree/main/ui/.ai/skills/skill-creator
Command: npx skills add https://github.com/noartem/kawa --skill skill-creator-noartem

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 Creator helps teams design and improve Claude skills by guiding end-to-end workflow: drafting SKILL.md, validating it, running iterative evals, benchmarking, and description optimization.

Core Features & Use Cases

  • Capture Intent: define what the skill should enable and when it should trigger.
  • Interview and Research: gather edge cases, determine dependencies, and plan tests.
  • Write SKILL.md: fill in name, description, compatibility, and references.
  • Run Eval Loop: execute tests, collect results, and iterate using evals and scripts as needed.
  • Benchmark and Analyze: aggregate timing and pass rates to compare variants.
  • Description Optimization: refine the frontmatter to improve triggering accuracy.
  • Outputs: provide structured evals.json, benchmarks, and viewer artifacts.

Quick Start

Define the skill's goal, create SKILL.md content, add any resources under scripts/, references/, and assets/, then run the evaluation loop to iterate until triggering becomes reliable.

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 tune Claude skills through iterative evaluation?

To create and tune Claude skills, define the skill's goal, draft the SKILL.md file with frontmatter, and run an iterative eval loop using scripts to test, collect results, and benchmark until triggering becomes reliable.

How do I improve triggering accuracy for Claude skill descriptions?

Improve Claude skill triggering accuracy by applying description optimization techniques to refine the frontmatter, then validate the changes by running benchmark tests to compare pass rates across variants.

What is the best way to benchmark and analyze Claude skill performance?

Benchmark Claude skill performance by aggregating timing and pass rates from the evaluation loop, then analyze the structured evals.json output and viewer artifacts to compare skill variants and identify improvements.

Do I need pyyaml to automate Claude skill workflow evaluation?

Yes, pyyaml is required as a dependency to automate Claude skill workflow evaluation, as it processes the YAML frontmatter structure used within the SKILL.md files for skill configuration and description optimization.

How does the evaluation loop work when refining a skill?

The evaluation loop works by executing tests against drafted SKILL.md content, collecting pass rate results, and allowing you to iterate using scripts and evaluative prompts to continuously improve skill triggering and performance.

What should be included in a SKILL.md file to ensure proper skill validation?

A SKILL.md file should include the skill name, description, compatibility details, and references to ensure proper validation, alongside any supporting resources organized under scripts, references, and assets directories.