Show HN: I built a mongo-express alternative using AI and almost zero coding

1 point

a year ago

I wanted to see how far AI could go in building a complete application, so I tasked Claude with creating a MongoDB admin interface similar to mongo-express. I only did quick code reviews and minor improvements, probably around 95% of the code was written by the AI.

For the AI, I used Anthropic Claude Sonnet 3.5. The result is MongoMan (https://github.com/aientech/mongoman), a modern MongoDB admin interface built with Next.js 15.1.2, shadcn/ui, and TypeScript. It features database stats, collection management, document CRUD operations, and index management. It's also containerized and ready to deploy.

The interesting part isn't just that AI has write the code, but how it handled the entire development process: suggesting the tech stack, implementing features incrementally, fixing bugs, and even writing the documentation.

However, although the current version works, I'm now maintaining and improving it myself as the project will grow. I guess this shows both the potential and current limitations of AI-assisted development.

The code is open source, and I've kept the commit history to show the development process. Curious to hear your thoughts on this approach to development and mongoman itself.

Cheers and happy new year!