AI agents are becoming the new users of the web, and most CMSes are invisible to them. What if Drupal could speak the language of AI agents natively? This session shows you how to expose Drupal as an MCP server, turning your content infrastructure into something AI agents can reason about, query, and act on, in production.

Prerequisite

- Familiarity with Drupal's REST or JSON:API modules
- No prior MCP experience required

Target Audience

Backend and full-stack Drupal developers and solution architects who are building or evaluating AI-assisted content workflows, and who want to future-proof their Drupal infrastructure for a world where AI agents are active participants in content systems, not just sideline tools.

Outline

The Model Context Protocol (MCP) is rapidly becoming the standard interface layer between AI agents and external systems. It is how tools like Claude, Cursor, and autonomous coding agents connect to APIs, databases, and services in a structured, safe, and composable way.

Most Drupal sites are built for human browsers. But the next wave of users will be AI agents: automated pipelines, LLM-powered assistants, and agentic workflows that need to read, create, and manage structured content. Drupal's entity system, typed fields, REST and JSON:API layers make it uniquely well-suited to serve this role, but only if you bridge it deliberately.
This session walks through the architecture and implementation of a Drupal MCP server: exposing content types, taxonomy, media, and workflows as MCP tools and resources that AI agents can discover and use. We will cover the protocol fundamentals, how to map Drupal's data model onto MCP primitives, authentication and permission scoping, and where the sharp edges are in production.

You will leave with a concrete architectural blueprint and the mental model to make any Drupal site agent-ready, not as a marketing claim, but as a working system.

Learning Objectives

- Understand what MCP is and why it matters for CMS platforms specifically

- Map Drupal's entity and field model onto MCP tools and resources

- Design authentication and permission scoping for agent access

- Identify the production pitfalls of exposing a CMS to autonomous agents

- Walk away with a reusable architectural blueprint for an MCP-enabled Drupal site

Experience level
Intermediate