Open Lovable: clone any webpage into a React app
An open-source sample app from the Firecrawl team that scrapes a webpage and hands it to an LLM to rebuild as a runnable React app, with conversational iteration, a choice of four models, and cloud sandbox previews—handy for replicating a frontend or spinning up a prototype fast.
Project facts
GitHub Ecosystem- License
- MIT
- Language
- TypeScript
- Stars
- 28,569
- Data checked
- 2026-09-25
Snapshot figures reflect the check date and may change over time.
Open Lovable is an open-source sample project from the team behind the Firecrawl scraping service, and it does one thing well: give it a website URL, and it uses Firecrawl to fetch the page content, then hands it to an LLM of your choosing to rewrite as modern React code, installing dependencies and running it in a cloud sandbox for a live preview. The project itself is built on Next.js and TypeScript. For frontend developers who see a site they like and want to riff on it, it collapses the cloning process down to a few chat turns. GitHubDaily recommended the project in August 2025, shortly after it launched.
Core features
- Clone straight from a link: paste the URL of the site you want to replicate, and the project calls Firecrawl to scrape the page, then has the AI rebuild it as a React app—no manual structure analysis or scaffolding required.
- Conversational development: if the output isn’t quite right, keep chatting—ask in plain language to adjust the layout, colors, or components, closing in on what you want turn by turn.
- Four models to choose from: the code-generation engine supports Gemini, Claude, OpenAI, or Groq, so you aren’t locked into a single vendor.
- Cloud sandbox preview: generated apps run in Vercel Sandbox by default, or you can switch to E2B; installing dependencies, starting Vite, and viewing the preview all happen in the cloud, keeping your local environment clean.
- Faster edits (optional): add a Morph API key and the AI’s code edits come back faster.
- Export and take it with you: projects generated in the sandbox can be packaged and exported so you can keep developing them in your own environment.
Typical use cases
- Use a site you like as a reference to clone a frontend skeleton, then build on it as the starting point for your own React project.
- Rapid prototyping: turn an existing page into a running React demo in minutes for reviews or discussions.
- Studying the source to learn engineering practice: the whole scrape-generate-sandbox pipeline is in the repo, useful for anyone researching how similar AI coding tools are built.
Quick start
- Clone the repo and install dependencies:
git clone https://github.com/firecrawl/open-lovable.git && cd open-lovable
pnpm install
-
In
.env.local, fill in the requiredFIRECRAWL_API_KEY, then pick one ofGEMINI_API_KEY,ANTHROPIC_API_KEY,OPENAI_API_KEY, orGROQ_API_KEY; the sandbox defaults to Vercel, but you can switch to E2B with the corresponding key. -
Start the dev server and open
localhost:3000in your browser:
pnpm dev
Summary
This suits developers who want to quickly turn an existing webpage into React code, teams that need a runnable prototype, and learners studying AI webpage-cloning pipelines. It’s licensed under permissive MIT, but the external dependencies aren’t trivial: a Firecrawl key is required, you need one of four model keys, and the sandbox needs a separate Vercel or E2B account—the setup cost before it runs isn’t small. The maintainers are also clear that this is a sample app; the full hosted commercial offering lives at their own product, Lovable.dev. Also worth noting: the repo’s last code push was mid-November 2025, more than ten months before this article was written (September 2026), so evaluate its maintenance status before adopting it.