MCP Server & Agent Toolkit

Description

MCP Server v2 and the dotCMS Agent Toolkit connect Claude, Cursor and other AI agents to dotCMS, so developers and technical content teams can describe the outcome they want while the agent does the work inside your guardrails.

The MCP Server gives an agent two tools, search and execute, to find the right dotCMS API endpoints and run scripts against them in an isolated sandbox. It can create content types, build pages, place content and move files, and dedicated tools confirm that a page actually renders.

The Agent Toolkit adds skills that teach any compatible agent how dotCMS works, including one that plans and builds a complete site.

The MCP Server was first released in September 2025 and was updated in October 2026 with a new architecture, the dotCMS Agent Toolkit, and one-command setup.

Get started

You need Node.js 20 or later and a dotCMS user + password (or API token). In a terminal on the machine where your editor or agent runs, go to your project's root folder and run:

npx dotcms agent setup

Answer three questions:

1. Your dotCMS instance URL

2. How to authenticate

3. Which code editors to connect.

Setup validates the token, adds the dotCMS MCP server to each code editor's configuration, installs the dotCMS skills, and starts the server to confirm it responds. It works with Claude Code, Cursor, VS Code with Copilot, Codex, Antigravity, Devin and OpenCode. No dotCMS instance handy? Try the public demo site at demo.dotcms.com with the user: admin@dotcms.com and password: admin.

Then prompt the agent a first task: "Give me a summary of all the assets in the dotcms instance, then create a content type called Event with a title, a start date and a location.". Setup options and manual configuration are in the MCP Server README.

Benefits

For Developers and Platform Teams

• Go from nothing to a connected agent in one command, with no hand-edited config files. Setup mints and verifies an API token, merges the dotCMS server into your editor's existing configuration and installs the skills.

• Let agents scaffold the repetitive work: content models, templates, containers, pages and themes, in either VTL-rendered or headless delivery.

• Keep the agent's context small. The agent discovers endpoints on demand instead of loading hundreds of tool definitions, and file transfers never pass through the model.

• Build on the same runtime. The @dotcms/ai package that powers the MCP server is available for developers building their own agents.

For Content Operations and Technical Content Teams

• Describe a change in plain language and have the agent find every affected item and make the edit, instead of working through the admin one item at a time.

• Ask questions of your content: stale pages, references to a retired product, content types with missing fields.

• Move drafts and external content into dotCMS content types without copying and pasting between tools..

For Architects and Security Teams

• No separate AI permission model. Every call runs as the user the API token belongs to, under that user's roles and permissions. An agent cannot do what that user could not.

• The API token stays on the MCP server and is never exposed to the code the agent writes. Sandbox code cannot make network requests of its own.

• The role is the boundary. The server steers agents to a curated set of endpoints so they find the right calls quickly, and the permissions of the user behind the token decide what an agent can do on every endpoint.

Use Cases

Stand up a new site section from one brief

A developer describes a new section, a content type, a listing page and a detail page, and points the agent at the team's design system. The agent creates the content type, builds the pages to match the design system, places the content and then verifies each page renders, flagging any empty slot so it gets fixed before anyone sees it.

Bulk edits from a single prompt

A content operations lead needs a product name changed across a large site. The agent searches for every affected item, shows the list, applies the edits through the same workflow actions an editor would use, and reports what changed.

Apply a design system to a new or existing site

A front-end developer points the agent at a design system. The site-building skill interviews them, writes a plan and a design file, builds the site, and keeps fixing and re-checking until every page type renders. Themes and assets upload to dotCMS in one step.

Give a partner team safe access

An architect creates an API token for a user with only the permissions a project needs. The agent works within those limits, so a consulting team can use an AI agent on one site without gaining access to others.

Resources

• MCP Server & Agent Toolkit Intro Blog Post

• dotCMS MCP Server README on GitHub

• MCP Server Documentation

• Intro to the MCP Server Blog Post

• Building with the MCP Server Video

• Developer Blog Post: MCP Server Context for AI

• MCP Server Feature Intro Video