Model Context Protocol

dbt can now function as a Model Context Protocol (MCP) server, providing a standardized framework that enables users to seamlessly integrate AI applications with their data warehouses. This ensures a consistent and efficient way to connect, access, and utilize data across various AI tools.

The MCP server runs in dbt using a personal access token and is available to all users across Cloud CLI, API, the Discovery API, and Semantic Layer. It provides access to private APIs (as AI Agent tools), text-to-sql, and SQL execution.

Note that access to the CLI, API, Discovery and the Semantic Layer is limited depending on your plan type.

For more information on MCP, have a look at Get started with the Model Context Protocol (MCP).

Remote auth not supported

MCP does not currently support remote authentication. Stay tuned for further updates.

Prerequisites

Before using MCP, read the instructions for the dbt-mcp repository.

Further reading

• Refer to our blog on Introducing the dbt MCP Server for more information.