This project provides an in-browser PostgreSQL sandbox with AI-powered features, targeting developers who need to quickly prototype, test, or visualize database interactions without complex setup. It offers instant, unlimited Postgres instances running directly in the browser, enhanced by LLM capabilities for tasks like CSV import, report generation, charting, and diagramming.

How It Works

The core innovation is the use of PGlite, a WebAssembly (WASM) port of PostgreSQL, enabling fully functional Postgres databases to run client-side. Each database instance is isolated and leverages browser IndexedDB for persistent storage. This approach eliminates the need for remote database containers or WebSocket proxies, offering a lightweight and immediate development experience.

Quick Start & Requirements

• Install dependencies: npm i
• Start local Supabase stack: npx supabase start
• Configure .env.local files with Supabase URL/anon key, OpenAI API key, and KV (Redis) variables.
• Start local Redis containers: docker compose -f ./apps/web/docker-compose.yml up -d
• Run development server: npm run dev
• Requires Node.js, npm, Docker, and an OpenAI API key.

Highlighted Details

• In-browser Postgres via PGlite (WASM).
• AI assistance for data import, reporting, charting, and diagramming.
• Data persistence via IndexedDB.
• Monorepo structure with Next.js frontend and browser proxy.
• Potential for deployment to S3 and other platforms.

Maintenance & Community

The project was formerly known as postgres.new and is community-driven under the Supabase umbrella. Links to community channels or roadmaps are not explicitly provided in the README.

Licensing & Compatibility

Licensed under Apache 2.0, which permits commercial use and linking with closed-source projects.

Limitations & Caveats

The project is in active development, with some features like deployment to S3 noted as "soon." The README emphasizes the renaming from postgres.new to avoid misleading users about its official status.