MCP Toolbox for Databases

MCP Toolbox for Databases is an open source Model Context Protocol (MCP) server that connects AI assistants, development environments, and software applications directly to relational and non-relational database systems. Developed by Google, it functions as both an off-the-shelf build-time MCP server and a run-time framework for developing secure, production-ready database tools. Software engineers, data analysts, and AI application developers use it to inspect schemas, execute read-only queries, and safely interact with operational data directly from IDEs or agent workflows. The server supports a broad range of storage engines, including PostgreSQL, MySQL, BigQuery, AlloyDB, Cloud SQL, Spanner, Firestore, SQL Server, Oracle, MongoDB, Redis, Snowflake, and ClickHouse. It includes prebuilt database tools such as list_tables and execute_sql, alongside configuration structures in tools.yaml to define parameterized SQL queries, custom toolsets, and prompts. Built-in connection pooling, IAM authentication support, and OpenTelemetry observability ensure consistent performance and security during database operations.

Category: Databases & Data Stores

Tags: data-access, database-management, SQL

Visit MCP Toolbox for Databases

How to install and configure MCP Toolbox for Databases

To install and configure MCP Toolbox for Databases in your MCP client (such as Claude Desktop or Gemini CLI), follow these steps: 1. Open your client's configuration file (such as claude_desktop_config.json or mcp.json). 2. Add the server under mcpServers using the @toolbox-sdk/server package via npx: json { "mcpServers": { "toolbox-postgres": { "command": "npx", "args": [ "-y", "@toolbox-sdk/server", "--prebuilt=postgres", "--stdio" ] } } } 3. Set the required database connection environment variables according to your database target, or supply a custom configuration using --config tools.yaml. 4. Save the configuration and restart your MCP client.

What you can do with MCP Toolbox for Databases

• Inspecting database structures by running list_tables to retrieve table names, column layouts, and data types before executing analysis. - Executing read-only SQL queries directly from an MCP client to retrieve records and troubleshoot application data during development. - Defining parameterized, restricted SQL queries within a tools.yaml file to expose safe, reusable querying tools to autonomous agents. - Grouping specialized database operations into distinct toolsets so that agents only load permissions and tools relevant to specific tasks. - Integrating database interaction tools directly into LangChain, LlamaIndex, ADK, or Genkit applications using the official Toolbox SDKs.

Key facts

• Open Source
• https://github.com/googleapis/mcp-toolbox
• Databases & Data Stores
• data-access, database-management, SQL

Part of MCP Servers

How do I install MCP Toolbox for Databases?

You can run MCP Toolbox for Databases using npx with the @toolbox-sdk/server package. Add the configuration to your MCP client settings file such as claude_desktop_config.json or mcp.json, specify the target database with the prebuilt flag, and provide the necessary connection credentials via environment variables.

What can MCP Toolbox for Databases do?

The server connects MCP clients to enterprise databases. It provides prebuilt tools for listing schemas and executing read-only SQL queries, and serves as a framework to create custom parameterized queries, prompts, and toolsets using a tools.yaml configuration file.

Which MCP clients work with MCP Toolbox for Databases?

MCP Toolbox for Databases functions with any standard MCP-compatible client or development environment. Common examples documented by the project include Gemini CLI, Google Antigravity, Claude Code, Codex, and Claude Desktop.

What databases are supported by MCP Toolbox for Databases?

Supported data stores include PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, MongoDB, Redis, Elasticsearch, CockroachDB, ClickHouse, Couchbase, Neo4j, Snowflake, Trino, and Google Cloud services like AlloyDB, BigQuery, Cloud SQL, Spanner, and Firestore.

Is MCP Toolbox for Databases open source?

Yes, MCP Toolbox for Databases is an open source project released under the Apache 2.0 license. The source code, documentation, and SDKs are maintained publicly on GitHub in the googleapis/mcp-toolbox repository.