Open Lovable

Open Lovable is a revolutionary open-source AI tool that transforms any website into clean React/Next.js applications in seconds. Built by Mendable AI as a free alternative to Lovable.ai, it uses advanced web scraping via Firecrawl and large language models to analyze target websites and regenerate their UI as modern React/TypeScript code with Tailwind CSS styling. The tool is MIT-licensed and completely free to use, modify, and self-host.

• AI-Powered Website Cloning uses cutting-edge AI models including Claude, GPT, and Groq to analyze and recreate any website as clean React components in seconds, leveraging Firecrawl for comprehensive web scraping and HTML/CSS analysis.

• Clean React/TypeScript Code generates production-ready React/Next.js applications with modern hooks, full TypeScript support, and Tailwind CSS styling for consistent, maintainable output.

• 100% Open Source & Free provides complete freedom with MIT licensing, allowing developers to fork, modify, and self-host freely without subscription fees, usage limits, or hidden costs.

• Secure Code Execution runs generated code safely in E2B Sandbox environment, ensuring secure and isolated testing and development workflows.

• Fully Customizable gives complete control over the generation process, allowing modification of prompts, swapping of AI models, and extension of frameworks to suit specific needs.

• Interactive AI Chat enables real-time refinement of results with natural language commands, letting users adjust styling, layout, or functionality instantly through conversational interaction.

• Multiple AI Model Support integrates with Claude, GPT-4, Groq/Kimi, and Google Gemini, providing flexibility in choosing the best model for specific use cases.

To get started, clone the repository from GitHub, install dependencies with npm, configure your API keys for E2B Sandbox, Firecrawl, and your preferred AI model, then run the development server. Simply paste any website URL into the chat interface and watch as the AI analyzes the site and generates clean React code. The tool typically achieves 80-95% visual accuracy and allows iterative refinement through natural language commands.