MengTo/Skills

Agent skills for designers and builders using Codex, Claude, Cursor, and other AI coding agents

What it solves

It addresses the inconsistency and "one-off" nature of AI coding agent outputs. Instead of relying on vague chat prompts, it provides a library of versioned, procedural "skills" that agents can load to execute complex, repeatable workflows for web design, game development, and UI implementation.

How it works

The project organizes prompts and workflows into a standardized folder structure. Each skill contains a SKILL.md file—a plain-language playbook that tells an agent exactly when to use the skill, the specific steps to follow, default settings to apply, and common pitfalls to avoid. These files are designed to be loaded as context into AI agents like Codex, Claude Code, and Cursor.

Who it’s for

Designers and developers who use AI coding agents to build high-end user interfaces, playable browser games (Three.js), and automated frontend workflows.

Highlights

  • Portable Skill Format: Uses a strict folder contract (SKILL.md, REFERENCES.md, ARTICLE.md) making workflows easy for agents to ingest.
  • Diverse Library: Includes over 100 skills across categories like Web Design (WebGL, GSAP, Tailwind), Game Development (Three.js, ARPG systems), and Codex operational workflows.
  • Reference-Driven: Emphasizes using screenshots and videos as prompts (e.g., "Video to Super Prompt") rather than long text descriptions.
  • Agent Agnostic: Compatible with multiple agent stacks including Codex, Claude, and Cursor.

Related

  • Project
  • Project
  • Project
  • Project
  • Project