Engineer's Must-Have! dbx Lightweight Database Tool + MCP Server Lets AI Agents Directly Operate on Data; A Cross-Platform, Open-Source, Free Lightweight Database Management Tool: dbx

1. Introduction

dbx is a database client tool that is open-sourced on GitHub and free to use. Developed in Rust, it runs with high execution efficiency. Unlike dBeaver, DataGrip, and Aqua Data Studio, which all require the Java Runtime, dbx only needs a single executable file to run. Its convenience of use is unmatched. Those who frequently work with databases are highly encouraged to give it a try.

dbx has built-in AI Chat integration. Once configured, you can directly use the current connection information as context to converse with a configured AI Provider, effectively generating the SQL statements you need.

If a portable.dbx file exists in the same folder as the dbx executable, it will automatically enter portable mode, creating a dbx.db file in the data subdirectory to store connection settings and other configuration data.

This time, I will first walk through the steps to establish a connection, then demonstrate how to use dbx to query and maintain .json files and .csv files—querying text file contents directly with SQL statements. The examples use Northwind data, and the test data can be found and downloaded by searching on GitHub.

2. UI Layout

The top area is the toolbar
The left side is the connection navigation panel
The center is the query editor and data Grid
The right side contains various property panels, including the AI chat area
Most functions are driven by right-click context menus

3. AI Settings

Using OpenCode as an example:

Provider: Select OpenAI Compatible
API Key: Paste in your API Key
Endpoint: https://opencode.ai/zen/go/v1
Model: First click Refresh on the right side, then select a model
Click Test at the bottom to verify the connection is working

+ Info

dbx has a built-in MCP Server. AI Agents can perform SQL operations on connections through dbx's MCP API.

4. Establishing a Connection

Using SQLite's Northwind.db as an example.

Click New Connection in the upper-left corner
Choose the database type you want to use, select SQLite, then click Next > in the lower-right corner

Enter a name and specify the file, then click Save & Connect

Right-click the connection → New Query to create a new query tab
You can also click a table to open the data browser tab

5. Creating a Connection with .csv Files

Using multiple Northwind .csv files as an example.

+ Tip

The DuckDB database can directly read and write .csv and .json files.

Click New Connection in the upper-left corner
Choose the database type you want to use, select DuckDB, then click Next > in the lower-right corner
Name: Give it a name. Color: Choose an identifying color
File Path: Click Create DuckDB File on the right to create a file in a specific folder (custom name; the default is database.duckdb)
- You can also enter :memory: to create an in-memory temporary database
Click Save & Connect

5.1. Single Table

-- List all file names
SELECT * FROM glob('j:/temp/Northwind/csv/*.csv');

-- Query file data
select * from 'j:\temp\Northwind\csv\Customers.csv';

5.2. Creating Tables with AI

Click AI in the upper-right corner and enter the following prompt to generate SQL statements that create tables from .csv files
If execution fails, use Fix with AI to repeatedly correct until successful, or switch to a different model

Read j:\temp\Northwind\csv\*.csv and generate a SQL statement for each .csv file in the format: create table filename as select * from 'full file path';