Describe a web app idea in plain English and get a working full-stack prototype with a database and auth generated for you.
Fork the Chef codebase to build your own AI app generator powered by a different database or AI model.
Study how an AI agent loop interprets natural-language prompts and calls language models to generate application code.
| get-convex/chef | remaxjs/remax | thunderbird/thunderbolt | |
|---|---|---|---|
| Stars | 4,566 | 4,568 | 4,562 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | hard | moderate | hard |
| Complexity | 4/5 | 3/5 | 4/5 |
| Audience | vibe coder | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Requires a Convex project, an OAuth app on the Convex dashboard, and API keys for at least one AI model provider such as Anthropic or OpenAI.
Chef is an AI-powered tool that lets you describe a web application in plain language and have it built for you automatically, including the backend database, user authentication, file uploads, and real-time features. It is made by the team behind Convex, an open-source database designed for web app development, and the connection to Convex is what makes the AI-generated code functional out of the box rather than requiring extensive manual setup afterward. The easiest way to use Chef is through the hosted website, which has a free tier. You describe what you want, Chef generates a working full-stack application, and you can start using or customizing it right away. The codebase in this repository is the open-source version of that hosted product, so developers who want to understand how it works, contribute bug fixes, or run their own version locally can do so. Chef started as a fork of another open-source AI app builder called bolt.diy. The team took that foundation and built on top of it, wiring everything into Convex's own APIs for data, authentication, and background tasks. The repository is organized into several folders: one for the browser-side UI components, one for the AI agent loop that interprets your prompts and calls language models, one for the database layer, and one for the template that every new project starts from. If you want to run Chef locally, you need to set up a Convex project, create an OAuth application on the Convex dashboard, add your own API keys for whichever AI models you want to use (such as Anthropic, OpenAI, or Google), and then run both the frontend and the Convex backend at the same time. The README walks through each step. One thing to note is that the current authentication system is specific to Convex's internal infrastructure, so forks intended for separate production deployments would need to replace it.
An AI-powered app builder that turns a plain-English description into a working full-stack web app with a database, authentication, and real-time features, powered by Convex, open-sourced so you can run or extend it yourself.
Mainly TypeScript. The stack also includes TypeScript, Convex, React.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.