R2 Data2


r2-data2

r2-data2 is a web-based SQL query tool, data browser, and visualization platform built with a Rust (Axum) backend and a React (TypeScript/Vite) frontend. The whole code (95%) is built by AI agent. As a human, I only generate the initial design and the instructions for the AI agent. There are several issues AI agent (using gemini-2.5-pro) keep failing:

• The initial tailwind 4 integration, even I give it the doc, it still use the old way.
• papaya::HashMap::pin() can't be used by async functions. Everytime the UI want's to remove it and use a different solution (mutex hashmap), so I have to manually fix it by using pin_owned().
• SqlEditor: when it needs to pull type import { TableColumn, DataTable } from '@sqlrooms/duckdb';. It kept using the wrong package, and I have to manually fix it.

This is still imperfect, but it's working and it just spent me 2-3 nights to build it (about 10 hours which I'm still doing several more projects/explorations while waiting for the AI agent to finish). So it is pretty amazing, especially for someone like me who knows only a little bit about react frontend development.


Features

Backend (Rust / Axum)

• Database Support: Connects to PostgreSQL and MySQL databases (via sqlx).
• Web Framework: Built with Axum, providing asynchronous request handling.
• Schema Introspection: API endpoint (/api/schema) to fetch detailed database, table, and column information (including constraints like PK/FK), with caching (moka).
• Query Execution: API endpoint (/api/execute-query) to run SQL queries against the selected database.
• JSON Results: Query results are returned efficiently as JSON.
• Configuration: Uses the config crate for managing database connection strings and other settings (config/default.toml).
• Static UI Serving: Embeds and serves the compile