Large Language Models like ChatGPT and Claude are increasingly used for search, support, dev tools, and content creation. Out of the box, they can read but not interact with your Drupal sites content and features. Model Context Protocol (MCP) solves this by defining a standard way for LLMs to interact with external data sources and tools. Teaching the LLM to interact with Drupal and perform searches, flag nodes, and more.

By the end of this session, you will be able to:

* Understand what MCP is and the problems it solves.
* Use MCP with Drupal’s JSON:API, entity system, and permissions to support retrieval-augmented generation (RAG) and action execution.
* Allow an LLM to authenticate and access member-only content and features.
* Implement a Node.js MCP server to broker requests between an LLM and Drupal.
* Build practical use cases, such as retrieving upcoming events, drafting blog posts from recent content, or generating Drupal modules and code explanations.

I’ll illustrate by exposing Drupal docs and Drupalize.Me tutorials to an LLM for more context when generating Drupal code. The focus is on learning how to configure your Drupal site—and its unique data and features—so an LLM can understand and interact with it.

This technical session is for developers interested in AI \+ Drupal integration, focusing on building custom MCP servers.