A 20 MB Open‑Source DB Client That Handles 80+ Databases, Replacing DBeaver and Navicat

Why DBX?

Traditional database clients such as DBeaver (Java‑based, slow startup, high memory) and Navicat (paid subscription) require separate tools for Redis, MongoDB, ClickHouse, etc., leading to fragmentation. DBX targets a lightweight, all‑in‑one, free solution: a single ~20 MB executable that supports more than 80 databases and is released under the Apache‑2.0 license.

20 MB Single File – Appearance

DBX runs without a Java runtime and does not bundle Chromium, resulting in cold‑start times measured in seconds. The UI follows a familiar modern client layout: a left‑hand connection tree, a top bar for creating connections, new queries, data transfers, and driver management, a central CodeMirror 6 query editor, a data grid with right‑click sorting, filtering, inline batch editing, and a right‑side SQL snippet library.

Supported Databases (80+)

Relational : MySQL, PostgreSQL, SQLite, Oracle, SQL Server, MariaDB, TiDB, plus domestic options such as 达梦, OceanBase, openGauss, KingBase.

Cache / Document : Redis, MongoDB.

OLAP / Time‑Series : ClickHouse, DuckDB, Doris, StarRocks, InfluxDB, TDengine.

Search : Elasticsearch, Meilisearch.

Extended Connectors : Snowflake, BigQuery, Neo4j, etc.

In a typical project where MySQL stores business data, Redis provides caching, and ClickHouse powers reports, developers previously needed three separate clients; DBX consolidates them into a single window.

Core Features

CodeMirror 6 Query Editor : syntax highlighting and auto‑completion.

Virtual‑Scrolling Data Grid : handles millions of rows without lag and supports inline editing.

ER‑Diagram Schema Visualization : displays table relationships.

Data Migration, Import/Export .

Redis / MongoDB Dedicated Browsers : key‑value and document views that do not require SQL thinking.

AI SQL Assistant

The AI is embedded directly in the query editor and can generate, explain, optimize, and fix SQL statements.

Supported models include Claude, OpenAI, and local Ollama, allowing teams to run models on‑premises so SQL never leaves the internal network.

AI‑generated SQL passes a built‑in safety check before execution.

Native MCP Server

AI agents such as Claude Code, Cursor, or Windsurf can reuse the connections already configured in DBX to query and manipulate data without re‑entering connection strings. npx @dbx-app/mcp-server When coding, an agent can automatically connect to a test database, generate the appropriate SQL, and return results within the development context.

Installation

Desktop builds are available for macOS, Windows, and Linux, with additional Docker Web and CLI options.

# macOS
brew install --cask dbx

# Windows
winget install t8y2.dbx
scoop install dbx

# Docker Web (port 4224)
docker run -d -p 4224:4224 -v dbx-data:/app/data t8y2/dbx:latest

Linux also offers a Flatpak package. Teams can run a Docker instance internally and share a web‑based DBX instance across the organization.