
MCP Toolbox for Databases
A solid choice when you need AI agents to talk directly to databases without writing connection boilerplate. This server gives you two modes: prebuilt generic tools for instant schema exploration and SQL execution against PostgreSQL, MySQL, BigQuery, Spanner, MongoDB, and a dozen other databases, or a YAML-driven framework for building custom, production-safe tools with structured queries and semantic search. The prebuilt tools work immediately with any MCP client through npx or Docker. The custom tools framework lets you define restricted queries in a tools.yaml config, which matters when you're moving past demos into real systems. It handles connection pooling, IAM auth for Google Cloud databases, and OpenTelemetry tracing out of the box. Four SDKs available if you want to integrate with ADK, LangChain, or custom agents.
- GitHub stars
- 16.9K
- tools
- 2
- transport
- HTTP
- last updated
- Nov 13, 2026
Tools
2 toolsPublic tool metadata for what this MCP can expose to an agent.
list_tablesPrebuilt generic tool for exploring database tables.
Prebuilt generic tool for exploring database tables.
No parameter schema in public metadata yet.
execute_sqlPrebuilt generic tool for executing SQL statements.
Prebuilt generic tool for executing SQL statements.
No parameter schema in public metadata yet.
Sponsors
Readme
Details
- Repository
- github.com/googleapis/mcp-toolbox
- Website
- github.com/googleapis/mcp-toolbox
- Categories
- Databases, AI & LLM Tools
- Package
us-central1-docker.pkg.dev/database-toolbox/toolbox/toolbox:1.17.0- Transport
- HTTP
- Registry
- MCP registry, active