skill-creator

Create and refine Claude Skills through iterative eval and trigger testing.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/TungND2k2/Milo-ai --skill skill-creator-tungnd2k2
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/TungND2k2/Milo-ai/tree/main/skills/skill-creator
Command: npx skills add https://github.com/TungND2k2/Milo-ai --skill skill-creator-tungnd2k2

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

This Skill provides an end-to-end workflow for creating and refining Claude Skills. It guides you through drafting SKILL.md, defining evals, running trigger evaluations, analyzing results, and iterating until the skill performs as intended.

Core Features & Use Cases

  • Draft and refine SKILL.md frontmatter and body, including naming, description, and compatibility notes.
  • Define and manage eval sets (evals.json) with prompts and expectations.
  • Run trigger evaluations across a pool of prompts, capture results, and generate summaries.
  • Iterate by updating descriptions, running new evals, and comparing improvements.
  • Benchmark and package skills for distribution (including .skill packaging).

Quick Start

Draft your SKILL.md, define an eval set, run the full eval/improve loop with the provided scripts, and review results to iterate.

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 refine Claude Skills for reliable triggering?

To create and refine Claude Skills, you draft the SKILL.md frontmatter, define an eval set in evals.json, run trigger evaluations, and iterate on descriptions until the skill achieves desired triggering accuracy and reliability.

What is the best way to evaluate and iterate on Claude Skill descriptions?

The best way to evaluate Claude Skill descriptions is to run trigger evaluations across a pool of prompts, capture results, generate summaries, and compare improvements after each iteration to ensure reliable performance.

How do I draft SKILL.md frontmatter and manage eval sets for prompt-engineering workflows?

You draft SKILL.md frontmatter including naming, descriptions, and compatibility notes, then define and manage eval sets by specifying prompts and expectations in an evals.json file for structured prompt-engineering workflows.

Do I need pyyaml to run trigger evaluations and package Claude Skills?

Yes, pyyaml is a required dependency to run the provided scripts that orchestrate the end-to-end workflow, including trigger evaluation, performance analysis, and .skill packaging for distribution.

Can I benchmark and package skills for distribution after iterating on trigger accuracy?

Yes, after iterating and achieving the desired trigger accuracy, you can benchmark the performance and package the finalized Claude Skills into a .skill file format for distribution.

Why does my Claude Skill fail to trigger correctly across different prompts?

A Claude Skill fails to trigger correctly when its description lacks precision; you resolve this by running trigger evaluations, analyzing performance summaries, and iteratively updating the SKILL.md description to improve accuracy.