The introduction of Model Control Protocol (MCP) enables AI models, such as Large Language Models (LLMs), to autonomously decide which tools to use in order to accomplish a task by providing context. The Couchbase MCP Server is an open-source offering that allows LLMs to perform actions against a Couchbase cluster through a well-defined set of tools. This server can be self-managed or hosted on Capella, and currently supports various tools such as fetching scopes and collections for buckets, running SQL++ queries, retrieving documents, inserting documents, deleting documents, and updating documents. To get started with the MCP Server, users can leverage existing clients like Claude desktop or Windsurf, and a video demonstration shows how to use the server in action, including verifying results by running SQL++ queries directly on the cluster. The development of new tools and updates is ongoing, with support for resources and prompts planned for the future.