skill-creator

Guide interviews, draft SKILL.md, generate eval prompts, and run evaluation loops.

1.3k|126|Updated Jan 8, 2026
One-click install
npx skills add https://github.com/poco-ai/poco-claw --skill skill-creator-poco-ai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: skill-creator
Source: https://github.com/poco-ai/poco-claw/tree/main/backend/assets/skills/skill-creator
Command: npx skills add https://github.com/poco-ai/poco-claw --skill skill-creator-poco-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Drafting, evaluating, and refining Claude Skills is labor-intensive. This Skill guides users through building SKILL.md, composing eval prompts, running iterative improvements, and preparing for a final upload.

Core Features & Use Cases

  • Automates the Skill authoring loop: generate descriptions, set up evals, run iterative improvements, and compare variants.
  • Supports end-to-end skill lifecycle: drafting, evaluation, optimization, packaging, and review submission.
  • Provides a path to publish: upgrade descriptions for triggering, run tests, and upload/publish the final Skill.

Quick Start

Use this Skill to draft a new skill, create eval prompts, run the improvement loop, and upload the finished skill.

Frequently Asked Questions about skill-creator

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

FAQPage Schema
How do I automate Claude Skill creation and iterative refinement?

Automate Claude Skill creation by guiding interviews, drafting SKILL.md files, generating eval prompts, and orchestrating evaluation loops. The workflow emphasizes repeatable testing, versioned iterations, and ready-to-upload benchmark results to iteratively refine skills.

What is the best way to evaluate and benchmark custom Claude Skills?

Evaluate and benchmark custom Claude Skills by generating eval prompts and running orchestrated evaluation loops. The process runs benchmarks, suggests improvements, and compares versioned iterations to optimize skill performance before publishing.

What is the end-to-end workflow for developing and publishing a Claude Skill?

The end-to-end workflow for developing a Claude Skill covers drafting, evaluation, optimization, packaging, and review submission. It upgrades descriptions for triggering, runs tests, compares variants, and guides publishing the final skill for review.

Do I need YAML to author and iterate on Claude Skills?

Yes, YAML is a required dependency to author and iterate on Claude Skills. The scripts use YAML to structure the automated skill creation, evaluation prompts, and versioned iteration configurations throughout the development lifecycle.