skill_creator

Create and refine YiYi skills with automated evaluation loops.

49|2|Updated Mar 20, 2026

Search Tags:#automation #workflow #evaluation #packaging #iteration #skill

Ready to run

npx skills add https://github.com/vibeinging/YiYi --skill skill-creator-vibeinging

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: skill_creator
Source: https://github.com/vibeinging/YiYi/tree/main/app/src-tauri/skills/skill_creator
Command: npx skills add https://github.com/vibeinging/YiYi --skill skill-creator-vibeinging

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Skill creation and iteration can be error-prone and slow. This Skill provides a structured, repeatable workflow to create new YiYi skills, test them with automated evals, and refine the description to improve triggering accuracy.

Core Features & Use Cases

End-to-end Skill creation: set up SKILL.md frontmatter, wire optional resources (scripts/references/assets), and validate with quick tests.
Evaluation-driven iteration: run trigger evaluation, generate benchmarks, and use analyzer/improve_description to refine prompts and triggers.
Packaging and deployment: package the skill into a .skill bundle after validation, and store evaluation artifacts for future audits.

Quick Start

Create a new skill folder named skill_name, add SKILL.md, then start the evaluation loop to iterate on the description and tests.

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 iterate skills with automated evaluation loops?▼

You can create and iterate skills by extracting your workflow into a SKILL.md file, then running automated train and test evaluation loops to validate trigger accuracy and refine descriptions.

What's the best way to package a validated skill for deployment?▼

Package a validated skill by bundling the SKILL.md and its wired resources into a .skill file, which stores evaluation artifacts and ensures the skill is ready for deployment audits.

How does evaluation-driven iteration improve skill trigger accuracy?▼

Evaluation-driven iteration improves trigger accuracy by running benchmark tests and using an analyzer to refine the description prompts, ensuring the skill activates reliably for intended workflows.

Do I need yaml to set up skill frontmatter and resources?▼

Yes, yaml is required to define the frontmatter configuration in SKILL.md and to wire optional resources like scripts, references, and assets during the initial skill setup.

What are the limitations of manually refining skill descriptions without automated evals?▼

Without automated evals, manual refinement is slow and error-prone, lacking structured benchmark generation and analyzer feedback to consistently validate trigger performance across iterations.