dbhub
DBHub, brought to you by Bytebase, is a zero-dependency, token-efficient MCP server that acts as a lightweight gateway between MCP clients (Claude Desktop, Claude Code, Cursor, VS Code, Copilot CLI) and databases. It supports PostgreSQL, MySQL, SQL Server, MariaDB and SQLite.
It exposes a small set of MCP tools: execute_sql, which runs SQL queries with transaction support and safety controls, search_objects, which searches and explores schemas, tables, columns, indexes and procedures with progressive disclosure, and custom tools, which are reusable parameterised SQL operations you define in a dbhub.toml configuration file. A built-in web workbench lets you run queries and custom tools and view requests visually. The README includes quick-start and multi-database setup sections, and the project is MIT licensed.
Features
Why choose it
Trade-offs
Maintenance check
- Last commit
- 2026-10-21
- Commits, last 90 days
- 95
- Releases, last 90 days
- 12
- Latest release
- v1.5.0 (2026-10-09)
- Open issues
- 2
- Recent issues with at least one comment
- 6/16
Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.
Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.
Most discussed open issues
- #146 👉👉👉 Please share your use case to shape DBHub roadmap25 comments · 2025-11-26
- #469 Feature: per-source `[sources.session]` settings applied on each new connection1 comments · 2026-10-09
Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.
Who it fits
- You want one MCP server for several engines: PostgreSQL, MySQL, SQL Server, MariaDB and SQLite.
- You care about token use: the README describes a small tool surface of execute_sql, search_objects and custom tools.
- You want to expose only reviewed, parameterised SQL by defining custom tools in dbhub.toml.
- You need MongoDB, Redis or another non-SQL store; only the five engines above are supported.
- You want to give the model free-form SQL on production data with a powerful account; execute_sql runs whatever SQL it is given.
- You need Google Cloud or Azure specific features such as IAM authentication.
How it differs
Both put several SQL databases behind one MCP server and both can restrict an agent to predefined tools. DBHub is lighter (MIT, zero dependencies, five engines, a built-in web workbench). MCP Toolbox from Google covers far more sources, including Google Cloud databases, and adds connection pooling, IAM authentication and OpenTelemetry, at the cost of YAML configuration.
The MySQL-only server is simpler still, but DBHub adds other engines, a schema-search tool and custom tools instead of arbitrary SQL only.
Compatibility
Quick start
Use cases
Alternatives
Featured in these lists
Comments
No comments yet. Be the first!