Google Cloud introduced the MCP Toolbox for Databases at Next 25 to enable AI agents to access enterprise data easily and securely.
The Toolbox, formerly known as Gen AI Toolbox for Databases, is an open-source MCP server facilitating connections between AI agents and various databases.
Toolbox simplifies development with reduced boilerplate code, enhances security with OAuth2 and OIDC, and offers observability with OpenTelemetry integration.
It provides scaffolding for implementing production-quality database tools, making them accessible to any client in the MCP ecosystem.
MCP Toolbox supports the Agent Development Kit (ADK) for building sophisticated multi-agent systems with precise control over agent behavior.
ADK allows developers to shape agent interactions, enable bidirectional audio and video streaming, and choose models and deployments that suit their needs.
Developers can deploy agents built with ADK using Vertex AI Agent Engine, streamlining the path from development to enterprise-grade deployment.
LangGraph support offers persistence layer for resilient, stateful agents, leveraging managed databases like AlloyDB for PostgreSQL and Cloud SQL for PostgreSQL.
By utilizing the langchain libraries, developers can save and load agent execution states seamlessly, ensuring reliability and manageability.
The article provides resources for getting started with Toolbox and MCP, such as a quick start guide, documentation, and GitHub repositories.