Why DBX Makes Navicat Obsolete: A Lightweight, AI‑Powered Database Tool

Problem with existing database clients

DBeaver bundles a several‑hundred‑megabyte Java runtime and starts slowly; Navicat requires a separate installation for each database type and carries a high license cost.

DBX overview

DBX is a 20 MB binary with no external dependencies. It runs directly on macOS, Windows, and Linux and supports more than 70 database systems, including MySQL, PostgreSQL, Redis, MongoDB, DuckDB, and Chinese platforms such as 达梦 and OceanBase.

Architecture

The application is written in Rust and built with Tauri. Its editor component uses CodeMirror 6, eliminating the Electron‑based Chromium layer and providing native‑level performance.

AI SQL assistant

Within the editor, after selecting a table the user can describe the desired operation in natural language. An AI model (switchable among Claude, OpenAI, Ollama, etc.) generates the corresponding SQL statement, which is first passed through a safety‑check step before execution, preventing malformed queries from affecting the database.

Installation

macOS: brew install --cask dbx Windows: use the provided installer package.

Linux: install via Flatpak, AppImage, or run the Docker image:

docker run -d -p 3000:3000 ghcr.io/t8y2/dbx:latest

MCP Server

The Docker image also starts a self‑hosted MCP Server that pre‑configures database connections for AI coding tools such as Claude Code, Cursor, and Windsurf, allowing those agents to query data without manual SQL authoring.

UI features

SQL syntax highlighting and intelligent code completion.

Execute queries with Cmd+Enter.

Persistent query history.

Nine selectable themes with automatic dark‑mode adaptation.

Repository

Source code:

https://github.com/t8y2/dbx