🗄️

SQLite Explorer

About SQLite Explorer

A fully offline, browser-based explorer for SQLite databases. Drop a .sqlite file onto the page and browse its tables, run SQL, inspect structure and view records — nothing is uploaded anywhere.

Your data never leaves your machine

This is the core promise of the app. The file you drop or pick is opened by a SQLite engine compiled to WebAssembly that runs inside your browser, in a web worker. Queries read pages straight from the local file on demand. There is no backend, no account, no telemetry endpoint — the app works with the network cable unplugged (the only exception being the optional AI model download described below).

What it is good at

  • Instantly opening databases of any size — multi-gigabyte files included
  • Exploring tables, structure and individual records read-only
  • Writing and running arbitrary SQL with an editor that knows your schema
  • Formatting messy SQL with one keystroke
  • Letting AI agents browse the database through WebMCP tools, with every action mirrored in the UI

It is a read-only explorer: the database opens in read-only mode and your file is never modified. For editing, use your normal SQLite tooling.

How it works

  • wa-sqlite (WebAssembly SQLite) runs in a web worker; a custom blob VFS serves page reads directly from the dropped file, so memory use stays at a few megabytes regardless of file size.
  • Monaco (the VS Code editor) provides the SQL editor, loaded lazily so the landing page stays instant.
  • @tanstack/react-virtual renders only the grid rows in view.
  • WebLLM + Qwen2.5-Coder-1.5B power optional ghost-text completions, running on your GPU via WebGPU. The model is served from a dedicated CDN (cdn.lucasschirm.com) and cached by your browser; on machines without WebGPU it simply stays off.

Open source

The source lives at github.com/lucasschirm/sqlite-explorer, and the app stands on the shoulders of these open-source projects:

  • sqlite-erd — interactive Entity-Relationship Diagrams for SQLite, powering the Diagram view (/visualize).