Important: This documentation covers Yarn 1 (Classic).
For Yarn 2+ docs and migration guide, see yarnpkg.com.

Package detail

@volter-ai-dev/flow-editor

volter-ai11.9kMIT1.11.3TypeScript support: included

AI Flow Editor - A visual node-based editor for AI workflows with ReactFlow, Subscribe.dev and Replicate API integration

react, reactflow, ai, workflow, visual-editor, node-editor, subscribe-dev, replicate, umd

readme

AI Flow Editor

A visual node-based editor for creating AI workflows with ReactFlow and Subscribe.dev integration.

Quick Start

# Install dependencies
bun install

# Start development server
bun run dev

Then open http://localhost:8000/index.html in your browser to see the main AI Flow Editor application.

Features

  • Visual Workflow Editor - Drag-and-drop interface for creating AI workflows
  • Multiple AI Models - Support for image generation, video generation, and text models
  • Node-based Connections - Visual connections between inputs, models, and outputs
  • User Authentication - Integrated Subscribe.dev authentication and billing
  • Real-time Execution - Execute workflows with live progress tracking
  • Generation History - Track all previous AI generations with persistent history
  • Professional UI - Clean interface with Phosphor icons and proper aspect ratios
  • UMD Format - Works in any JavaScript environment with no external dependencies
  • Programmatic Initialization - Load predefined workflows via window.FLOW

Installation

npm install @volter-ai-dev/flow-editor

Prerequisites

Before using the Flow Editor, make sure you have the following dependencies loaded:

<!-- Process polyfill -->
<script>
  window.process = { env: { NODE_ENV: 'development' } };
</script>

<!-- React -->
<script crossorigin src="https://unpkg.com/react@18/umd/react.development.js"></script>
<script crossorigin src="https://unpkg.com/react-dom@18/umd/react-dom.development.js"></script>

<!-- Babel for JSX transformation -->
<script src="https://unpkg.com/@babel/standalone/babel.min.js"></script>

<!-- ReactFlow -->
<script src="https://unpkg.com/reactflow@11.11.4/dist/umd/index.js"></script>
<script>
  // Load ReactFlow CSS dynamically
  fetch('https://unpkg.com/reactflow@11.11.4/dist/style.css')
    .then(response => response.text())
    .then(css => {
      const style = document.createElement('style');
      style.textContent = css;
      document.head.appendChild(style);
    });
</script>

<!-- Subscribe.dev -->
<script src="https://unpkg.com/@subscribe.dev/react@0.0.161/index.umd.js"></script>

<!-- Tailwind CSS (for styling) -->
<script src="https://cdn.tailwindcss.com"></script>

Note: The Flow Editor includes all necessary icons internally - no external icon libraries are required.

Usage

ES Modules / React App

// Note: This package is primarily designed for UMD usage
// ES module support may be limited in the current version
import FlowEditor from '@volter-ai-dev/flow-editor';

function App() {
  return (
    <FlowEditor 
      projectToken="your-subscribe-dev-project-token"
    />
  );
}

export default App;

UMD / Script Tag

<!DOCTYPE html>
<html>
<head>
  <!-- Prerequisites (see above) -->

  <!-- Flow Editor -->
  <script src="https://unpkg.com/@volter-ai-dev/flow-editor@latest/dist/flow-editor.umd.js"></script>
</head>
<body>
  <div id="flow-editor-container"></div>

  <script type="text/babel">
    // FlowEditor is available as window.FlowEditor
    const { ReactFlowProvider } = window.ReactFlow;
    const { SubscribeDevProvider } = window.SubscribeDevReact;

    // Get FlowEditor from compiled UMD
    const FlowEditor = window.FlowEditor.default;

    // Auth wrapper with providers
    const AuthWrapper = () => {
      return (
        <ReactFlowProvider>
          <FlowEditor />
        </ReactFlowProvider>
      );
    };

    // Main App with SubscribeDevProvider
    const App = () => {
      return (
        <SubscribeDevProvider baseUrl="https://api.subscribe-dev.com">
          <AuthWrapper />
        </SubscribeDevProvider>
      );
    };

    const root = ReactDOM.createRoot(document.getElementById('flow-editor-container'));
    root.render(<App />);
  </script>
</body>
</html>

Programmatic Initialization

You can initialize the Flow Editor with predefined nodes and edges by setting window.FLOW before the component mounts:

<script>
  window.FLOW = {
    nodes: [
      {
        id: '1',
        type: 'textInput',
        position: { x: 100, y: 250 },
        data: { 
          prompt: 'A beautiful sunset over mountains',
          label: 'Text Input'
        }
      },
      {
        id: '2', 
        type: 'model',
        position: { x: 500, y: 100 },
        data: {
          model: 'black-forest-labs/flux-schnell',
          width: 1024,
          height: 1024,
          outputCount: 2,
          label: 'Model'
        }
      }
    ],
    edges: [
      {
        id: 'e1-2',
        source: '1',
        target: '2', 
        sourceHandle: 'prompt',
        targetHandle: 'prompt'
      }
    ]
  };
</script>

API Reference

FlowEditor Props

Prop Type Required Default Description
projectToken string No - Your Subscribe.dev project token (can be set via SubscribeDevProvider)

Note: The FlowEditor is typically wrapped with SubscribeDevProvider which handles authentication and API configuration.

Supported AI Models

The Flow Editor supports various AI models through Subscribe.dev:

Image Generation

  • Flux Schnell (black-forest-labs/flux-schnell) - Fast image generation
  • Flux Kontext (black-forest-labs/flux-kontext-max) - Image generation with image input
  • Flux Dev (black-forest-labs/flux-dev) - Higher quality image generation
  • Google Nano Banana (google/nano-banana) - Supports 0-2 image inputs

Video Generation

  • WAN Video (wan-video/wan-2.2-5b-fast) - Video generation from text/image
  • Seedance Lite (bytedance/seedance-1-lite) - High quality video generation
  • Seedance Pro (bytedance/seedance-1-pro) - Premium video generation

Text Generation

  • Models available through Subscribe.dev's text completion API

Workflow Components

Input Nodes

  • Text Input - For text prompts and instructions
  • Image Input - For image uploads and references

Model Nodes

  • AI Models - Execute AI generation with configurable parameters
  • Support for multiple outputs (1-8 per model)
  • Real-time progress tracking
  • Parameter controls (width, height, aspect ratio, etc.)

Connections

  • Visual connections between compatible node types
  • Automatic type validation
  • Support for multiple outputs connecting to multiple inputs

Key Features

Generation History

  • Track all previous AI generations automatically
  • Persistent history across sessions
  • Modal view with generation details
  • One-click access to previous results

Authentication & Billing

  • Integrated Subscribe.dev authentication
  • Credit usage tracking
  • Subscription management
  • Sign in/out functionality

Workflow Execution

  • Topological sorting for correct execution order
  • Parallel output generation
  • Real-time status updates
  • Error handling and retry logic

User Interface

  • Dark theme optimized for AI workflows
  • Professional Phosphor icon set
  • Proper aspect ratios for media previews
  • Keyboard shortcuts (Delete/Backspace to remove nodes/edges)
  • Hover controls for node management
  • Minimap and zoom controls
  • Clickable media outputs with copy functionality

Development

To work on this package locally:

# Clone the repository
git clone https://github.com/volter-ai/flow-editor.git

# Install dependencies  
bun install

# Start development server (builds and serves with file watching)
bun run dev

This will:

  1. Build the flow-editor UMD bundle
  2. Start webpack in watch mode (rebuilds on file changes)
  3. Start a local server on http://localhost:8000

Running the Main App

Once the development server is running, open your browser to:

http://localhost:8000/index.html

This is the main AI Flow Editor application with a pre-configured workflow demonstrating:

  • Text input nodes
  • AI model nodes (Flux Schnell for image generation)
  • Visual node connections
  • Real-time workflow execution
  • Multiple output generation

Other Available Commands

# Build once (without serving)
bun run build

# Serve existing build (without rebuilding)
bun run serve

# Build once and serve (no file watching)
bun run dev:serve

# Build for production
NODE_ENV=production bun run build

Additional Example Files

The local server also hosts other example implementations:

  • http://localhost:8000/editor_simple.html - Simple Subscribe.dev integration demo
  • http://localhost:8000/test-umd.html - UMD package functionality test
  • http://localhost:8000/code_editor.html - Advanced multi-model editor
  • http://localhost:8000/flowgraph.html - Flow graph visualization
  • http://localhost:8000/sheet.html - Sheet-based interface
  • http://localhost:8000/perms_experiment.html - Authentication experiments

Note: All examples require the UMD bundle to be built first. The bun run dev command handles this automatically.

Example Files

This repository contains several example HTML files demonstrating different aspects of the flow editor:

  • index.html - Main example showcasing the full flow editor with predefined workflow
  • editor_simple.html - Simple Subscribe.dev integration demo
  • test-umd.html - Basic UMD package functionality test
  • code_editor.html - Advanced multi-model editor with flow management
  • flowgraph.html - Flow graph visualization and editing
  • sheet.html - Alternative sheet-based interface
  • perms_experiment.html - Permissions and authentication experiments

Each file demonstrates different features and can be used as starting points for your own implementations.

Subscribe.dev Integration

This package requires a Subscribe.dev account and project token. Get started at subscribe.dev:

  1. Sign up for a Subscribe.dev account
  2. Create a new project
  3. Copy your project token
  4. Use it in the projectToken prop

License

MIT

Support

For issues and feature requests, please use the GitHub Issues page.