skills.rest
zinan92
zinan92COMMUNITY
Explore all skills by this author →
GitHub

skill-creator

Create and refine Claude Code skills with SKILL.md content and eval benchmarks.

1|Updated Apr 5, 2026
Search Tags:#skill creation#benchmarking#evaluation#grader#description optimization#iteration loop#triggering accuracy
One-click install
npx skills add https://github.com/zinan92/repo-evals --skill skill-creator-zinan92
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/zinan92/repo-evals/tree/main/repos/anthropics--skill-creator/runs/2026-05-05/run-static-checks/artifacts
Command: npx skills add https://github.com/zinan92/repo-evals --skill skill-creator-zinan92

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Helps you design a Claude Code skill from scratch or refine an existing one by creating a loop that drafts the skill, builds realistic test prompts, runs triggering and output evaluations, and then iterates until performance is measurable and auditable.

Core Features & Use Cases

  • Skill creation and iterative improvement: Convert user intent into a usable SKILL.md with progressively refined instructions and test coverage.
  • Quantitative evaluation and variance-aware benchmarking: Run evals, compute pass rates, and compare against baselines to see whether changes truly improve outcomes.
  • Description optimization for triggering accuracy: Optimize the frontmatter description so the skill triggers at the right times (and avoids false triggers) using a held-out test set.
  • Viewer and analyst workflow: Generate a review UI to collect qualitative feedback, plus post-hoc analysis to surface patterns behind wins/losses.

Quick Start

Use the skill-creator to create a new skill by describing what the skill should do, then generate test prompts and run evaluations to iteratively improve the SKILL.md until triggering and output quality meet your goals.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I evaluate and benchmark Claude Code skills against baselines?▼

You optimize skill triggering by refining the YAML frontmatter description and testing it against a held-out set. This process increases correct invocation rates while actively reducing false triggers for unrelated prompts.

What is the process for iterative skill creation from a basic idea?▼

Yes, you can diagnose false triggers and failed outputs by using analyzer-based diagnosis and viewer-based review. This workflow generates a review UI to collect qualitative feedback and surfaces patterns behind wins and losses.

Do I need YAML frontmatter to build and improve skills?▼

You can optimize skill triggering by refining the YAML frontmatter description and testing it against a held-out set. This process increases correct invocation rates while actively reducing false triggers for unrelated prompts.

Why does my Claude Code skill trigger incorrectly on unrelated prompts?▼

Yes, you can diagnose false triggers and failed outputs by using analyzer-based diagnosis and viewer-based review. This workflow generates a review UI to collect qualitative feedback and surfaces patterns behind wins and losses.

Can I use parallel workers to run skill evaluation prompts?▼

Iterative skill creation involves drafting a SKILL.md file from your idea, generating realistic test prompts, and running evaluations. You loop through this process, progressively refining instructions and test coverage until performance is measurable and auditable.

What's the best way to reduce false triggers when creating a new skill?▼

You optimize skill triggering by refining the YAML frontmatter description and testing it against a held-out set. This process increases correct invocation rates while actively reducing false triggers for unrelated prompts.

Related Skills

Explore More Product & Management Skills→
xlsamaxlsama

skill-creator

Create and evaluate AI skills with SKILL.md files and baseline tests.

Community
Advanced
vnmoorthyvnmoorthy
2

canary

Monitor live apps for console errors and performance regressions post-deploy.

Community
Advanced
man-navlakhaman-navlakha

skill-creator

Create and refine AI skills with SKILL.md files and automated testing.

Community
Advanced
clinecline
19

skill-creator

Create and iteratively improve AI agent skills through structured evaluation workflows.

Official
Advanced