Skip to content

AI-assisted development

ratatui-kit ships an AI agent skill — a packaged knowledge base that teaches your AI coding assistant the framework’s real components, props, and hooks, the element! macro, the input-layer system, the router, and the idioms for building polished TUIs. It uses the open skills convention, so it works with any agent that supports it. With it installed you can ask your assistant to “build me a terminal todo app” or “add a confirmation modal” and get code that compiles and follows the framework’s conventions, instead of guessed APIs.

Why it helps: ratatui-kit is a niche framework, so an assistant working from memory tends to hallucinate APIs — the wrong event hook, removed macro syntax, props that don’t exist. The skill gives it an accurate, source-derived reference to work from, plus a build workflow that ends in cargo check.

The skill lives in this repository under skills/ratatui-kit/. Add it with the skills CLI:

npx skills add yexiyue/ratatui-kit --skill ratatui-kit

This installs the ratatui-kit skill. Your assistant consults it automatically whenever you ask it to build or extend a ratatui-kit / terminal UI — you don’t have to mention the skill by name.

  • SKILL.md — the mental model, a from-zero build workflow, a minimal compiling app template, the element! essentials, input/state guidance, palette-neutral visual-polish principles, and a pitfalls checklist.
  • references/ — deeper references the assistant reads on demand: every built-in component with its exact props, every hook signature, the macro grammar, the events / state / routing model, the full theming system (every Palette/*Theme field, writing a theme-aware custom component), and idioms for building polished UIs.

Just describe what you want to your assistant:

  • “Scaffold a new ratatui-kit todo app with a text field, a j/k list, and a delete confirmation.”
  • “Add a second screen to my app using the router.”
  • “Make this modal capture keyboard input so the background list stops reacting while it’s open.”

The assistant reads the skill, uses the correct APIs and idioms (the element! macro, #[component], use_state / use_event_handler, input layers, the routes! router), and — following the skill’s guidance — verifies the result compiles before handing it back.