skill-creator

Guide teams through drafting, validating, and iterating Claude skills.

1|1|Updated Feb 4, 2026
One-click install
npx skills add https://github.com/brianlan/improved-ai-agent --skill skill-creator-brianlan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/brianlan/improved-ai-agent/tree/main/skills/skill-creator
Command: npx skills add https://github.com/brianlan/improved-ai-agent --skill skill-creator-brianlan

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill guides users through creating new skills, refining their descriptions, and measuring skill performance using an end-to-end plan-build-validate workflow. It covers capturing intents, drafting SKILL.md, generating evaluation prompts, running trigger evaluations, and iterating based on results to improve triggering accuracy and operator flow.

Core Features & Use Cases

  • Capture Intent: Understand user goals, define skill scope, and draft an initial SKILL.md.
  • Interview & Research: Proactively ask edge-case questions to shape requirements and metrics.
  • Write the SKILL.md: Produce a well-structured entry with a YAML frontmatter and inline operational instructions.
  • Evaluation Loop: Generate eval prompts, run trigger tests, and compile results for iteration.
  • Iterate & Improve: Use evaluation feedback to refine prompts, tooling, and descriptions, establishing a repeatable improvement loop.

Quick Start

Draft an initial SKILL.md, generate 3–5 trigger-eval prompts, and run the end-to-end plan-build-validate loop to start iterating.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I build and evaluate Claude skills to ensure reliable triggering?

To build Claude skills, draft a SKILL.md with YAML frontmatter and inline instructions, generate 3–5 trigger-eval prompts, run trigger tests, and iterate based on evaluation results to refine descriptions and improve triggering accuracy.

What is the plan-build-validate workflow for prompt engineering?

The plan-build-validate workflow captures user intents, drafts a structured SKILL.md, generates evaluation prompts, runs trigger tests, and uses compiled results to iterate on prompt engineering and improve operator flow.

How do I write a SKILL.md file with proper frontmatter for Claude?

Write a SKILL.md file by structuring it with YAML frontmatter and inline operational instructions, capturing intents and edge-case requirements through proactive interviews to shape metrics and skill scope.

Can I use this skill-creator workflow to refine descriptions and measure skill performance?

Yes, you can refine descriptions and measure skill performance by generating evaluation prompts, running trigger evaluations, and using feedback to iterate on prompts, tooling, and descriptions within a repeatable improvement loop.

Do I need pyyaml to generate evaluation prompts and run trigger tests?

Yes, pyyaml is a required dependency to parse YAML frontmatter and support the evaluation loop that generates trigger-eval prompts and compiles results for iteration.

Why are my Claude skills not triggering reliably during operator flow?

Unreliable triggering often stems from poorly defined descriptions or frontmatter; running trigger-eval prompts and iterating based on compiled evaluation results refines descriptions and ensures skills trigger reliably.