mcp-toolbox
MCP Toolbox for Databases, from Google, is an open-source MCP server that connects AI agents, IDEs and applications to databases. It was originally called Gen AI Toolbox for Databases and was renamed to reflect its MCP focus. It serves two purposes: a ready-to-use MCP server with prebuilt generic tools such as `list_tables` and `execute_sql`, and a framework for defining custom, restricted tools for production agents.
It suits two audiences. Developers can connect Gemini CLI, Claude Code, Codex, Antigravity or another MCP client to a database and explore data in natural language during development. Teams building agents can define structured queries, semantic search and NL2SQL tools in a YAML configuration, so the agent only runs the operations you allow. Supported sources include Google Cloud databases (AlloyDB, BigQuery, Cloud SQL, Spanner, Firestore) and many others such as PostgreSQL, MySQL, SQL Server, Oracle, MongoDB, Redis, Elasticsearch, ClickHouse, Neo4j and Snowflake. Connection pooling, IAM authentication and OpenTelemetry are built in, and SDKs integrate it with ADK, LangChain and LlamaIndex.
Getting started: install with Homebrew (`brew install mcp-toolbox`), Go, Docker or `npx @toolbox-sdk/server`. Use `--prebuilt=postgres` (or another database) for instant generic tools, or pass `--config tools.yaml` with your own tool definitions.
Features
Why choose it
Trade-offs
Compatibility
Quick start
Use cases
Alternatives
Featured in these lists
Comments
No comments yet. Be the first!