中文

Cognot is an open-source, flexible, and extensible AI Workflow Engine focused on AI image generation and video processing. It empowers users to create, execute, and manage complex AI-driven workflows through an intuitive visual interface.

• December 11, 2025 🙌: Support for HunyuanVideo and Wan 2.2 Video Models
• December 08, 2025 🎠: Added Context Search Node & Optimized User Experience
• December 07, 2025 🎊: Supported Import of ComfyUI-Format Workflows
• December 01, 2025 ✨: Enabled GPU Acceleration for NVIDIA Graphics Cards

🚧 Under Development 🚧

Cognot is currently in active development. We are working diligently to implement all planned features and continuously improve the engine's stability and performance.

• Python: 3.8+
• Node.js: 16+
• npm: 8+
• Operating System: Windows, macOS, Linux

• Stable Diffusion Integration: Generate images from text prompts using Stable Diffusion models
• Image-to-Image Transformation: Convert images based on text prompts with precise control
• Advanced Parameter Control: Multiple samplers, resolution options, and configuration settings
• Custom Model Support: Load and use your preferred models for specialized results

• Comprehensive Video Handling: Load and process various video formats
• Frame-by-Frame Analysis: Extract and analyze individual video frames
• Optical Flow Calculation: Advanced motion analysis for smooth video transformations
• Wan2.2 Integration: Leverage powerful video processing capabilities

• Visual Drag-and-Drop Designer: Intuitive interface for creating complex workflows
• Real-time Execution Monitoring: Track progress and performance visually
• Import/Export Functionality: Share and reuse workflows across projects
• Version Control: Manage workflow versions and history

• Extensible Architecture: Easily create and integrate custom nodes
• Built-in AI Task Nodes: Ready-to-use nodes for common AI operations
• ComfyUI Adapter: Seamlessly use nodes from the ComfyUI ecosystem
• Type Safety: Advanced type checking for reliable node connections
• Cross-Platform Compatibility: Consistent behavior across operating systems

• Parallel Execution: Optimize performance with parallel branch execution
• Smart Queue Management: Efficient resource utilization and task scheduling
• Robust Error Handling: Comprehensive error recovery mechanisms
• Detailed Logging: Debug and monitor with extensive execution logs

• Responsive Web Interface: Access from any modern browser
• Dark/Light Theme Support: Comfortable viewing in any environment
• Intuitive Node Panel: Search and filter nodes with ease
• Real-time Visualization: Dynamic workflow progress tracking

• Stable Diffusion v1-5: runwayml/stable-diffusion-v1-5
• Stable Diffusion XL: stabilityai/stable-diffusion-xl-base-1.0
• Stable Diffusion 3: stabilityai/stable-diffusion-3-medium
• Stable Diffusion 3.5: stabilityai/stable-diffusion-3.5-large
• Stable Cascade: stabilityai/stable-cascade
• Flux: black-forest-labs/FLUX.1-dev

• Stable Diffusion v1-5 (Image-to-Image): runwayml/stable-diffusion-v1-5

2. Hunyuan Series

• Text-to-Image
• Hunyuan DiT: Tencent-Hunyuan/HunyuanDiT-v1.2

3. Qwen Series

• Qwen-VL-Chat: A dialogue model that supports mixed image-text input
• Text-Only
• Qwen Text: A text-only dialogue model ##Model Features
• Stable Diffusion Series: Widely adopted open-source diffusion models that support diverse image generation tasks
• hunyuan DiT: A diffusion model developed by Tencent, delivering high-quality image generation capabilities
• Qwen Series: Large language models developed by Alibaba. Qwen-VL supports image understanding and multimodal interaction
• All models have been integrated into the Model Cache Manager, which enables efficient model loading and memory management.

Cognot adopts a layered architecture designed for flexibility, scalability, and ease of extension:

• Graph Parser: Converts workflow definitions into executable graphs
• Execution Engine: Manages workflow execution, node scheduling, and resource allocation
• Node Registry: Central repository for node definitions and implementations
• Type System: Ensures type safety and compatibility between nodes

• API Gateway: RESTful endpoints for workflow management
• WebSocket Service: Real-time communication between client and server
• File Service: File upload/download and storage management
• Configuration Service: Application settings and configuration management

• React Frontend: Modern web interface built with Vite
• Context Providers: Global state and application context management
• Component Library: Reusable UI components for workflow design and management

1. Clone the repository:

   git clone https://github.com/CognotEngine/cognot.git cd cognot

2. Create and activate a virtual environment:

   # Windows python -m venv venv venv\Scripts\activate # macOS/Linux python3 -m venv venv source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

1. Navigate to the frontend directory:

   cd frontend

2. Install dependencies:

   npm install

# From project root directory python api/gateway/main.py

Backend API will be available at http://localhost:8000

# From frontend directory npm run dev

Frontend application will be available at the URL shown in the terminal output (typically http://localhost:3000)

Open your web browser and navigate to the frontend URL to start creating and executing workflows!

• Core engine code is located in the core/ directory
• API endpoints are defined in api/gateway/main.py
• Add new node types by creating new files in core/ and registering them with the @register_node decorator

• React application code is in frontend/src/
• Components are organized in frontend/src/components/
• Context providers are located in frontend/src/contexts/
• Run frontend in development mode:

  npm run dev

• Build for production:

  npm run build

Cognot features deep integration with ComfyUI, allowing seamless use of the extensive ComfyUI node ecosystem:

• Automatic Node Conversion: Convert ComfyUI nodes to Cognot-compatible format automatically
• Type System Mapping: Ensure type safety between different node systems
• Metadata Synchronization: Maintain consistent node descriptions and properties
• Seamless Execution: Use converted nodes directly in Cognot workflows

1. CheckpointLoaderSimple: Load a Stable Diffusion model
2. CLIPTextEncode: Encode text prompt into CLIP embeddings
3. KSampler: Generate latent space representation
4. VAEDecode: Decode into final image

Cognot supports GPU acceleration for AI model nodes on systems with compatible NVIDIA GPUs:

1. Ensure latest NVIDIA drivers are installed
2. Install CUDA and cuDNN (if required by AI models)
3. The system will automatically detect and use available GPUs for AI tasks

• Unified Platform: Combine image generation and video processing in one workflow
• Extensive Integration: Leverage both built-in nodes and ComfyUI ecosystem
• Developer-Friendly: Easy to extend with custom nodes and functionality
• Performance Optimized: Efficient execution engine with parallel processing
• Open Source: Community-driven development and transparency
• User-Centric Design: Intuitive interface for both beginners and experts

We highly welcome contributions from the community! If you'd like to contribute to Cognot:

1. Fork the repository
2. Create a new branch for your feature or fix
3. Make your changes and commit with descriptive messages
4. Push your changes to your fork
5. Submit a pull request

Cognot is released under the MIT License. See the LICENSE file for more information.

For questions, suggestions, or support, please open an issue on GitHub or join our development community.

⭐ Star this repository if you find Cognot useful!