Why DBX Can Replace Navicat: A 15 MB Open‑Source Client for 40+ Databases

First impression

I handed the 15 MB DBX installer to a colleague who constantly complained about Navicat’s price; it started in three seconds, connected to our test database, and he immediately dragged the Navicat icon to the recycle bin after seeing the smooth UI and the AGPL‑3.0 license.

Why existing clients fall short

During years of development I have tried more than five database clients. DBeaver is feature‑rich but slow to start because it loads a Java VM and consumes a lot of memory. Navicat is convenient but its commercial price hurts individual developers. TablePlus feels good but only runs on macOS. Moreover, projects often use MySQL, Redis, MongoDB, ClickHouse, TiDB, OceanBase, openGauss and other databases, forcing developers to switch tools constantly.

DBX’s look and feel

Opening DBX shows a clean layout: a left‑hand schema browser for quick navigation, a central CodeMirror 6 editor with syntax highlighting and smart completion, and a right‑hand result grid that uses virtual scrolling so tens of thousands of rows remain fluid.

15 MB, zero runtime dependencies

DBX is built on Tauri 2 + Rust (frontend: Vue 3 + TypeScript; backend: pure Rust database drivers). Unlike DBeaver, it does not require a Java runtime, Python environment, or embedded Chromium, making the download‑and‑run experience instant on Windows, macOS and Linux, and even as a self‑hosted Docker container.

Support for 40+ databases

DBX directly supports most databases developers encounter:

Relational : MySQL, PostgreSQL, SQLite, MariaDB, SQL Server, Oracle, CockroachDB

Document / key‑value : MongoDB, Redis (all data types)

Analytical : ClickHouse, DuckDB, Doris, StarRocks, SelectDB

国产信创 : TiDB, OceanBase, openGauss, GaussDB, KingBase, Vastbase, GoldenDB, 达梦 DM, 虚谷 XuguDB

Through JDBC it can also reach Snowflake, Trino, Hive, Neo4j, BigQuery and others, eliminating the need to juggle multiple tools.

AI SQL assistant

Select a table in the editor, describe the desired result in natural language, and the AI instantly generates the corresponding SQL. For example, typing “show the number of users registered in the last 7 days” produces a safe SELECT statement after a security check that prevents dangerous DROP or DELETE commands. The assistant works with Claude, OpenAI, or local models via Ollama, keeping data on the machine.

Data table details

The result grid uses virtual scrolling, allowing millions of rows without lag. Inline editing highlights changes, and batch saving is one click. Right‑click filtering (LIKE / NOT LIKE), column resizing, and export to CSV, JSON, Markdown, XLSX or INSERT statements are provided, similar to DataGrip.

ER diagram and schema diff

Right‑click a database or table to generate an ER diagram that visualizes primary‑foreign key relationships, with zoom and SVG export. The Schema Diff feature compares two connections, produces ALTER statements, and applies them with a single click—useful for migrations.

MCP integration

DBX implements the Model Context Protocol (MCP), allowing AI coding tools such as Claude Code, Cursor or Windsurf to read the database connections configured in DBX. Example configuration:

{
  "mcpServers": {
    "dbx": {"command": "npx", "args": ["-y", "@dbx-app/mcp-server"]}
  }
}

This lets AI assistants query databases and debug SQL without leaving the editor.

Dark mode and localization

Native dark mode follows the system title bar, and nine editor themes are available. The UI supports Chinese, English and Spanish, with automatic updates.

Installation

macOS (Homebrew): % brew install --cask dbx Windows (Scoop):

% scoop bucket add dbx https://github.com/t8y2/scoop-bucket
% scoop install dbx

Docker self‑hosted:

% docker run -d --name dbx -p 4224:4224 -v dbx-data:/app/data t8y2/dbx

After starting, open http://localhost:4224 in a browser; the container works on both amd64 and arm64. The binary can also be downloaded from GitHub Releases.

Conclusion

DBX combines a tiny footprint, extensive multi‑database support, AI‑assisted SQL, and MCP integration into a solid, free AGPL‑3.0 project. If you frequently work with many databases and want to avoid Java runtimes and commercial licenses, DBX is worth trying.