convez.ai
PluginDesenvolvimento WebIntermediário
DanWahlin

ZingIt

DanWahlin/zingit

Anote elementos UI e envie para IA implementar automaticamente

Ver no GitHub
BomPluginScore · de 0 a 100

Ferramenta de anotação visual para UI que permite marcar elementos da página e enviar alterações automaticamente para assistentes de IA (Claude, Copilot, Codex). Dispensa descrições textuais e screenshots manuais ao integrar visualmente as mudanças desejadas com geradores automáticos de seletores CSS.

Instalação

$npx convez-ai DanWahlin/zingit

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
7
Forks
1
Acompanhando
0
Saúde
57%
Último commit
há 2 meses
Versão
—
Issues abertas
0
Contribuidores
5

Casos de uso

  • Acelerar feedback de design em desenvolvimento web
  • Comunicar mudanças visuais a agentes IA sem screenshots
  • Iterar rapidamente sobre UI com implementação automática
  • Documentar bugs e melhorias visuais para assistentes de código

Veredito

ZingIt resolve um real ponto de fricção no workflow design-to-code, eliminando descrições verbosas e screenshots repetitivos. Bem pensado para integrar com o ecossistema de assistentes Claude/Copilot, mas requer setup local com uma das plataformas de IA suportadas.

README

A tradução para o português deste README ainda está na fila.

ZingIt

Features | Quick Start | Usage | Configuration | Troubleshooting

Point-and-click UI feedback that gets automatically implemented by AI

Tired of describing what element to change in the UI, taking screenshots, and copying/pasting them into your AI assistant? Streamline how you share UI changes! Select elements, mark them with the changes you want to make, and send directly to AI agents for automated fixes.

Point → Click → Describe → ZingIt to AI

Try the Live Demo →

Features

Visual Markers

Click any element on your page to add notes about changes you want. Smart CSS selectors are automatically generated to target the exact element.

Screenshot Capture

Automatically captures screenshots of marked elements to provide visual context to AI agents. No more manual screenshot taking!

Multi-Agent Support

Works with Claude Code, GitHub Copilot CLI, and OpenAI Codex. Choose your preferred AI assistant and watch it work in real-time.

Send Changes with a Click

Once you've marked your changes, simply click the sparkle icon (✨) in the ZingIt toolbar to send everything to your AI agent. Watch as it generates code updates live!

Additional Features

  • Real-time Streaming - Watch the AI work in real-time
  • Smart Selectors - Auto-generates CSS selectors for precise targeting
  • Change History - Track all modifications made by your AI assistant
  • Remote/Local Detection - Warns when editing published sites vs local development

Quick Start

Option 1: Clone the Repo and Run Locally

  1. Install an AI agent - Install and login to one of these AI agents on your dev machine:

  2. Clone and setup:

git clone https://github.com/danwahlin/zingit.git
cd zingit
npm install
  1. Start the ZingIt server and demo site:
npm run dev

This starts both the server (ws://localhost:3000) and the client dev server concurrently.

  1. Visit http://localhost:5200/?zingit to see ZingIt in action!

  2. You'll be prompted to select an AI agent. Start marking!

Option 2: Add to Your Website

  1. Start the ZingIt server:
npx cross-env PROJECT_DIR=/path/to/your/project npx @codewithdan/zingit

Replace /path/to/your/project with your project path.

  1. Add the ZingIt script to your HTML pages just before the closing </body> tag:
<script src="https://cdn.jsdelivr.net/npm/@codewithdan/zingit@latest/dist/zingit-client.js"></script>
  1. Run your website's dev server and visit it with ?zingit added to the URL.

Example: http://localhost:5200/?zingit

Usage

  1. Press Z to toggle mark mode on/off
  2. Click elements on your page to mark them - add notes about changes you want
  3. Click the sparkle icon (✨) in the toolbar to send markers to your AI agent
  4. Watch the agent work in real-time - the response panel shows streaming updates

Toolbar Icons

IconDescription
ON/OFFToggle mark mode
✨ (Sparkle)Send markers to AI agent
🕒 (Clock)View change history
📋 (Copy)Export markers as Markdown
🧹 (Broom)Clear all markers
?View keyboard shortcuts
⚙️ (Gear)Open settings
✕Close ZingIt toolbar

Keyboard Shortcuts

KeyAction
ZToggle mark mode on/off
Ctrl/Cmd+ZUndo last marker
?Show help overlay
`Toggle ZingIt visibility
EscClose current panel/modal
Ctrl/Cmd+EnterSave marker (in modal)

Configuration

Click the gear icon for settings:

SettingDefaultDescription
WebSocket URLws://localhost:3000Server connection
Project Directory(server default)Override project path
Highlight Color#fbbf24Element highlight color
Marker Color#3b82f6Marker badge color

Architecture

zingit/
├── client/          # Lit web components (browser UI)
│   ├── src/components/  # UI components
│   ├── src/services/    # WebSocket, storage
│   └── dist/            # Built bundle
└── server/          # WebSocket server + AI agents
    └── src/agents/  # Claude, Copilot, Codex integrations

Troubleshooting

WebSocket not connected

  • Ensure server is running: npx cross-env PROJECT_DIR=/path npx @codewithdan/zingit
  • Check WebSocket URL in settings (default: ws://localhost:3000)
  • Make sure you've added ?zingit to your URL

Agent not responding

  • Verify AI agent is installed and authenticated
  • Check server logs for error messages

Markers not persisting

  • Markers are URL-specific and stored in localStorage
  • Changing pages clears markers

Changes not appearing on published site

  • If you see a "Remote" badge in the toolbar, you're editing a published site
  • Changes are saved locally only - to see them, run the project locally or deploy the updated files
  • For best experience, use ZingIt with local development (localhost)

License

MIT


Made by Dan Wahlin | Issues | Contributing

Ver o original no GitHub

Bom para

Relacionados