WebDesignAgent is an autonomous agent designed to simplify website creation for users ranging from casual creators to developers. It transforms textual descriptions, uploaded images, or visual layout references into functional, multi-page websites, offering iterative refinement and human feedback integration for tailored results.

How It Works

The agent leverages a hybrid approach combining large language models (LLMs) for understanding user intent and generating content, with visual models for layout and image integration. It supports various input modalities, including text-to-website, image-to-website, and visual-clue-to-website, enabling complex designs through a combination of these. The system allows for multi-page generation, user-specific modifications, and iterative refinement based on feedback, aiming for a seamless and controlled website development process.

Quick Start & Requirements

• Install: git clone https://github.com/DAMO-NLP-SG/WebDesignAgent.git && cd WebDesignAgent && pip install -r requirements.txt
• Prerequisites: Python, API keys for supported LLMs (OpenAI, Claude, GLM, Qwen) and image generation models (DALL-E 3, CogView-3, SD3). Configuration is done via config.yaml.
• Usage: Supports both terminal (python webdesign.py) and GUI (python gui.py) modes.
• Docs: https://github.com/DAMO-NLP-SG/WebDesignAgent

Highlighted Details

• Supports multiple LLMs including OpenAI, Claude, GLM, and Qwen.
• Integrates with image generation models like DALL-E 3, CogView-3, and SD3.
• Enables creation of multi-page websites with dynamic redirects.
• Allows for user-specific page additions/deletions and iterative code refinement.

Maintenance & Community

The project is actively developed, with recent updates adding support for Qwen and SD3 models. A WeChat group is available for consultation.

Licensing & Compatibility

The repository does not explicitly state a license in the provided README. Users should verify licensing for commercial use or integration with closed-source projects.

Limitations & Caveats

The project is described as a "rough demo" with ongoing development. Qwen model integration has noted issues with asynchronous calls and slow generation speeds. Support for local code modification and backend code generation is listed on the TODO list.