One-click install
npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-skill-creator-jangtrinh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ck:skill-creator
Source: https://github.com/jangtrinh/demo-Lam/tree/main/.claude/skills/skill-creator
Command: npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-skill-creator-jangtrinh

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Skill creator automates authoring Claude skills and improving them with evaluation-driven iteration, so skills trigger reliably and follow the right instructions.

Core Features & Use Cases

  • Eval-driven iteration: test trigger accuracy, grade outcomes, compare with baselines, then optimize prompts and descriptions.
  • Progressive disclosure workflow: refine metadata first (so the skill activates), then improve the SKILL.md instructions and bundled resources.
  • Packaging and validation: validate SKILL.md frontmatter and structure before producing a distributable skill zip.
  • Benchmark reporting: aggregate results into benchmark.json/benchmark.md for faster analysis.

Quick Start

Use the skill-creator to create a new skill directory named "my-skill" and generate an initial SKILL.md and starter resources.

Frequently Asked Questions about ck:skill-creator

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

FAQPage Schema
How do I create Claude Code skills that trigger reliably?

To create Claude Code skills, you start by generating a new skill directory with an initial SKILL.md file and starter resources. The skill-creator workflow validates the required SKILL.md frontmatter fields, such as name and description, before packaging the skill for safe distribution.

What is eval-driven optimization for skill descriptions?

Skill benchmarking reports aggregate evaluation results into benchmark.json and benchmark.md files. These reports provide faster analysis by comparing trigger accuracy and graded outcomes against baselines, enabling continuous improvement of the skill's prompts and descriptions.

How do I package and validate a SKILL.md file for distribution?

You package and validate a SKILL.md file by checking its frontmatter for required name and description fields, verifying the directory structure, and bundling optional resources from scripts/, references/, and assets/ directories. This validation process ensures safe distribution of the skill zip.

Does skill creation require a specific directory structure?

Skill creation requires a valid root SKILL.md file with frontmatter containing name and description fields. Optional resources can be organized under scripts/, references/, and assets/ directories to support progressive disclosure, deterministic scripts, and bundled references within the skill package.

What is progressive disclosure in Claude skill workflows?

Progressive disclosure in Claude skill workflows is a refinement strategy that optimizes metadata first to ensure the skill activates, then improves the SKILL.md instructions and bundled resources. This structured approach ensures reliable triggering before adding complex processing logic.