Skip to main content

​
What is DBHub?

DBHub is a lightweight Model Context Protocol (MCP) server that bridges AI assistants with your databases. It implements the MCP specification to provide a standardized interface for database operations, enabling AI agents like Claude Desktop, Claude Code, Cursor, and VS Code to interact with your databases intelligently.
DBHub is brought to you by Bytebase, the open-source database DevSecOps platform.

​
Understanding MCP servers

The Model Context Protocol (MCP) is a standardized interface that allows AI assistants to interact with external data sources and tools. DBHub implements this protocol as an MCP server, acting as a gateway between:
  • MCP Clients: AI assistants (Claude Desktop, Claude Code, Cursor, VS Code, Copilot CLI)
  • Databases: Your PostgreSQL, MySQL, MariaDB, SQL Server, and SQLite databases
+------------------+    +--------------+    +------------------+
|                  |    |              |    |                  |
|  Claude Desktop  +--->+              +--->+    PostgreSQL    |
|                  |    |              |    |                  |
|  Claude Code     +--->+              +--->+    SQL Server    |
|                  |    |              |    |                  |
|  Cursor          +--->+    DBHub     +--->+    SQLite        |
|                  |    |              |    |                  |
|  VS Code         +--->+              +--->+    MySQL         |
|                  |    |              |    |                  |
|  Copilot CLI     +--->+              +--->+    MariaDB       |
|                  |    |              |    |                  |
+------------------+    +--------------+    +------------------+
    MCP Clients           MCP Server             Databases

​
Supported databases

DBHub supports five major database systems through a unified interface:

PostgreSQL

Full support for schemas, procedures, functions, and advanced features

MySQL

Complete MySQL compatibility with all standard operations

MariaDB

Native MariaDB connector with full feature support

SQL Server

Microsoft SQL Server support including named instances and NTLM authentication

SQLite

File-based and in-memory SQLite databases

​
Key features

DBHub is designed to work out of the box with no additional dependencies. Simply install via npm or run with Docker, and you’re ready to connect to your databases.
DBHub implements only two core MCP tools (execute_sql and search_objects) to maximize your AI assistant’s context window. The search_objects tool uses progressive disclosure, allowing you to start with minimal token usage (just object names) and request more details only when needed.
Connect to multiple databases simultaneously using TOML configuration. Perfect for managing production, staging, and development databases from a single DBHub instance.
Securely connect to remote databases through SSH bastion hosts with support for:
  • Password and key-based authentication
  • Multi-hop tunneling via ProxyJump
  • Automatic SSH config file parsing
  • Keepalive for stable long-running connections
DBHub includes a web-based interface for testing queries, running custom tools, and viewing request traces without requiring an MCP client.
Built-in safety controls include:
  • Read-only mode to prevent accidental modifications
  • Row limiting to prevent overwhelming responses
  • Query timeouts to prevent runaway operations
  • Per-tool configuration for granular control

​
MCP tools

DBHub exposes three types of MCP tools:

​
Built-in tools

execute_sql

Execute SQL queries with transaction support and safety controls. Supports multiple statements separated by semicolons.

search_objects

Search and explore database schemas, tables, columns, indexes, and procedures with progressive disclosure.

​
Custom tools

Define reusable, parameterized SQL operations in your dbhub.toml configuration file. Perfect for common queries that you want to expose as dedicated tools. Learn more about custom tools.

​
How it works

  1. Configure: Set up DBHub with your database connection details via command-line arguments, environment variables, or TOML configuration
  2. Connect: DBHub establishes a connection to your database(s) and registers MCP tools
  3. Discover: Your AI assistant discovers available tools through the MCP protocol
  4. Execute: The assistant can execute SQL queries, explore schemas, and invoke custom tools
  5. Respond: DBHub formats results and returns them to the assistant
DBHub supports both stdio transport (for desktop MCP clients) and HTTP transport (for network-based clients and the built-in workbench).

​
Next steps

Installation

Install DBHub using Docker, npm, or build from source

Quick start

Get started in minutes with Docker or demo mode

Configuration

Learn about multi-database setup and advanced configuration

Workbench

Explore the built-in web interface

Build docs developers (and LLMs) love

Get started for freeTalk to us