A 20 MB Open‑Source DB Client That Outperforms 1 GB DataGrip

Introduction

The author discovered dbx on GitHub Trending (10k+ stars) and decided to compare it with familiar tools such as DBeaver, Navicat, and DataGrip.

Overview

dbx is an Apache‑2.0 open‑source database client built with Tauri 2 (Rust backend) and Vue 3 + TypeScript (frontend). It uses CodeMirror 6 for the editor, offers native installers for macOS, Windows, and Linux, and provides a Docker‑based web version. The project claims no data collection.

Comparison Snapshot

Key differences include:

Price : dbx – free open‑source; DBeaver – free community; Navicat – paid; DataGrip – subscription.

Size : dbx – 15 MB; DBeaver – several hundred MB (requires Java); Navicat – >100 MB; DataGrip – ~1 GB.

Supported databases : dbx – 60+; DBeaver – many; Navicat – a dozen; DataGrip – mainstream.

AI capability : dbx – built‑in; DBeaver – plugin; Navicat – partial; DataGrip – requires AI subscription.

MCP support : dbx – native; others – none.

Multi‑Database Support

dbx can connect to MySQL, PostgreSQL, SQLite, Redis, MongoDB, ClickHouse, DuckDB, SQL Server, Oracle, TiDB, OceanBase, openGauss, GaussDB, DM, 人大金仓, TDengine, Qdrant, Milvus, Weaviate, Kafka, RocketMQ, Pulsar, and more.

SQL Editor

The editor (CodeMirror 6) provides syntax highlighting, schema‑aware auto‑completion, Cmd+Enter execution, selective execution, built‑in formatting, nine themes, query history, snippet library, and persistent tabs.

Integrated AI Assistant

Users can describe desired results in natural language to generate SQL, ask the assistant to explain queries, optimize slow queries, or fix errors. Generated SQL passes a safety check before execution. Supported models include Claude, OpenAI, Ollama, and any OpenAI‑compatible API such as DeepSeek, Kimi, or Tongyi.

MCP Server

dbx includes a native MCP server, allowing AI tools like Claude Code, Cursor, or Windsurf to reuse dbx‑configured connections without manually copying schema information.

CLI Tool

Install the command‑line client with npm install -g @dbx-app/cli and use it to run scripts via Codex.

Data Grid and Export

Results are displayed with virtual scrolling, editable cells, and a preview of the SQL that will be executed. Filters mirror DataGrip’s UI, and export options include CSV, JSON, Markdown, XLSX, and generated INSERT statements.

Schema Utilities

Features include ER diagrams, foreign‑key visualization, cross‑connection schema diff, visual execution plans, column‑level lineage, full‑text object search, and a change‑preview list before applying modifications.

Data Migration

CSV/Excel can be imported directly, cross‑database sync is supported with previewed diffs, and dragging Parquet/CSV/JSON files invokes DuckDB for instant preview without creating tables. DBeaver and Navicat connection configurations can be imported.

Redis and MongoDB Support

Redis offers pattern search, batch operations, TTL editing, and a command executor for all six data types. MongoDB provides document CRUD with pagination and supports Atlas and replica‑set connection strings.

Security Features

Both SSH key and password tunneling are supported, with separate proxies for database and AI traffic. The client auto‑reconnects, prompts confirmation for destructive actions, encrypts exported config files, and allows connection color‑coding (e.g., red for production).

Installation

macOS: brew install --cask dbx Windows (Scoop/WinGet):

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

Docker:

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

Access the web UI at http://localhost:4224.

Author’s Opinion

At 15 MB, dbx’s feature set is impressive, especially the native AI and MCP integration, which many competitors add only as after‑thought plugins. However, the project is single‑maintainer driven, with over 800 unresolved issues, so long‑term maintenance is uncertain. The author recommends keeping DataGrip as a fallback for critical production databases while using dbx as the primary client.

Open‑Source Repository

https://github.com/t8y2/dbx