Verified project record
t8y2/dbx
DBX is a ~25 MB multi-database client that connects to 100+ databases including MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, ClickHouse and SQL Server, with desktop, Docker, CLI, an optional AI SQL assistant, and a separate MCP server for AI coding agents.
Project overview
It packs support for 100+ databases into a single ~25 MB binary with no Java JRE, Python venv, or bundled Chromium, and adds an AI SQL assistant and an MCP server that lets AI coding agents use DBX-configured connections.
- Project type
- MCP · Data Processing
- Use cases
- Data Analysis · Coding & Development
- Deployment
- Refer to project documentation
- License
- Apache-2.0
Best for
- Developers and data teams who need to manage many different databases from one lightweight tool and prefer to avoid Java runtimes, Python virtual environments, or bundled Chromium.
- Users who want optional AI help writing, explaining, optimizing, or fixing SQL, with the flexibility to use a remote API model (Claude, OpenAI), a local model via Ollama, or any OpenAI-compatible endpoint.
- Developers who want AI coding agents to query databases through an MCP server using their existing DBX connection configuration.
Key capabilities
- Connects to 100+ databases in a single ~25 MB app; named examples include MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, ClickHouse, and SQL Server.
- Describe what you want in plain language and get SQL back; it can also explain queries, optimize SQL, and fix errors, with built-in safety checks applied to AI-generated SQL. Requires an optional model configuration (Claude, OpenAI, local models via Ollama, or any OpenAI-compatible endpoint).
- A separate Rust-powered MCP server lets AI coding agents query databases using DBX-configured connections; it runs via npx @dbx-app/mcp-server.
Limitations and risks
- AI-generated SQL is run through built-in safety checks, which implies a recognized risk of unsafe generated SQL; treat AI output with review before running against production data.
- The AI SQL assistant is optional and needs a model provider (Claude, OpenAI, Ollama, or an OpenAI-compatible endpoint); using remote providers sends requests outside your machine.
- GPU requirements, minimum hardware, coding requirements, and telemetry behavior are not documented; verify these against your environment before adopting.
Getting started
- Download the latest release from the Releases page, or install with brew install --cask dbx on macOS.
- Run docker run -d --name dbx -p 4224:4224 -v dbx-data:/app/data t8y2/dbx:latest, then open http://localhost:4224 in your browser. Setup difficulty is documented as easy with no listed blockers.
Alternatives and comparisons
- A ~30MB cross-platform database client built with Go and a native WebView, offering a schema-aware AI assistant and MCP agent integration across SQL, cache, vector, MQ, search and time-series sources, if DBX's coverage or footprint does not match your needs.
- If your focus shifts from database administration to consolidated financial data access across Python, workspace, MCP, and REST surfaces, OpenBB targets that problem instead.
Project comparisons
Evidence and sources
- GitHub project description: 25 MB lightweight cross-platform database client for 100+ databases, including MySQL, PostgreSQL, SQLite, Redis, MongoDB, DuckDB, SQL Server, and Dameng. Built-in AI, MCP Server,…
- README: MySQL, PostgreSQL, SQLite, Cloudflare D1, Redis, MongoDB, DuckDB, ClickHouse, SQL Server, Oracle, Elasticsearch
- README: No Java JRE. No Python venv. No bundled Chromium. DBX ships as a single small binary — download, install, connect.
- README: Describe what you want in plain language — get SQL back. DBX can explain queries, optimize SQL, fix errors, and run AI-generated SQL through built-in safety checks. Works with Cla…
- README: DBX provides a separate Rust-powered MCP server that lets AI coding agents query databases using connections configured in DBX.
Search
Find projects, verify facts, compare options, and filter by your requirements
Try a searchA click only fills the search box; you stay in control
Project Details
0