Agentic Factory (also known as Agent Fleet in the UI) is an interactive 3D visualization tool for designing, organizing, and managing AI agent architectures. Think of it as "Cloudcraft for agents" - a visual way to map out your agentic stack, from orchestrators and sub-agents to DSPy programs and Foundry resources.
- 3D Visualization: Immersive Three.js-based 3D scene for visualizing agent architectures
- AI-Powered Generation: Use Google Gemini AI to generate architectures from natural language prompts
- Multiple Node Types: Support for agents, tools, tasks, DSPy modules, strategies, phases, configs, surfaces, and annotations
- Provider System: Built-in support for Agent Framework, DSPy, and Foundry ecosystems
- Templates: Pre-built templates for common patterns (orchestrator, DSPy programs, Foundry stacks)
- Export/Import: Save and load your architectures in
.af.jsonformat - Interactive Editing: Drag-and-drop nodes, create connections, group nodes into zones
- Dark/Light Mode: Automatic theme switching based on system preferences
- Frontend: React 19, TypeScript
- 3D Rendering: Three.js via @react-three/fiber and @react-three/drei
- Styling: Tailwind CSS 4 with shadcn/ui components
- Build Tool: Vite 6
- AI Integration: Google Gemini API (@google/genai)
- State Management: React Hooks
- Node.js (v18 or higher recommended)
- npm or yarn
- Google Gemini API Key - Get one from Google AI Studio
-
Clone the repository:
git clone <repository-url> cd agentic-factory
-
Install dependencies:
npm install
-
Set up environment variables: Create a
.env.localfile in the root directory:GEMINI_API_KEY=your_api_key_here
-
Start the development server:
npm run dev
-
Open your browser to
http://localhost:3000
npm run build
npm run preview- Add Nodes: Use the sidebar to add different types of nodes (agents, tools, tasks, etc.)
- Create Connections: Click and drag from one node to another to create connections
- Edit Properties: Select a node to edit its properties in the right panel
- Use Templates: Click template buttons to add pre-configured architecture patterns
- AI Generation: Use the toolbar to generate architectures from natural language prompts
- Export/Import: Save your work as
.af.jsonfiles and import them later
- AGENT: AI agents with roles and system prompts
- TOOL: External tools and APIs
- TASK: Data processing tasks and workflows
- DSPY_MODULE: DSPy program specifications
- CONFIG: Configuration files and settings
- PHASE: Workflow phases and stages
- STRATEGY: Decision strategies and routing logic
- SURFACE: Grouping zones for visual organization
- ANNOTATION: Text labels for documentation
- dataFlow: Data passing between nodes
- controlFlow: Control and execution flow
- callsTool: Tool invocation
- retrievesFrom: Knowledge retrieval
- deploysTo: Deployment relationships
- dependsOn: Dependency relationships
agentic-factory/
├── components/ # React components
│ ├── Scene.tsx # 3D scene rendering
│ ├── Sidebar.tsx # Node creation sidebar
│ ├── Toolbar.tsx # Top toolbar
│ ├── PropertyPanel.tsx # Node property editor
│ └── ...
├── services/ # Business logic
│ ├── geminiService.ts # AI generation service
│ └── fileFormat.ts # Import/export handling
├── lib/ # Utilities
│ └── utils.ts # Helper functions
├── src/ # Styles
│ └── globals.css # Global styles
├── types.ts # TypeScript type definitions
└── App.tsx # Main application component
See docs/DEVELOPMENT.md for detailed development setup and guidelines.
npm run dev- Start development servernpm run build- Build for productionnpm run preview- Preview production build
Comprehensive documentation is available in the docs/ directory:
- Getting Started - Detailed setup instructions
- Architecture - System architecture overview
- API Reference - Service APIs and type definitions
- User Guide - Application usage guide
- Development Guide - Development environment setup
- Contributing Guide - Contribution workflow
We welcome contributions! Please see CONTRIBUTING.md for guidelines on how to contribute to this project.
For security concerns, please see SECURITY.md for our security policy and how to report vulnerabilities.
This project is licensed under the MIT License - see the LICENSE file for details.
- Built with Three.js for 3D rendering
- Powered by Google Gemini for AI generation
- UI components from shadcn/ui
For issues, questions, or contributions, please use the GitHub Issues page.