Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Automattic has added writing capabilities to WordPress.com’s MCP integration, giving AI agents like Claude and ChatGPT the ability to create posts, create pages, manage comments, and restructure content, all through a natural conversation, with human approval at every step.
For most of the last six months, connecting an AI agent to your WordPress.com site has meant giving it a window. You can ask Claude or ChatGPT questions about their content, check out site analytics, or check which posts haven’t been updated in a year. Useful, but fundamentally passive.
On Friday, Automattic added a door.
WordPress.com has launched writing capabilities for its Model Context Protocol (MCP) integration, which allows AI agents to create and modify content directly on your site.
The update adds 19 new operations across six content types: posts, pages, comments, categories, tags, and media. From a single natural language message, an agent can compose and publish a post, create a landing page using your theme’s blocking patterns, approve and respond to comments, reorganize category structures, or fix missing alt text throughout your media library.
The underlying architecture, MCP, an open protocol that standardizes how applications provide context to large language models, was first introduced on WordPress.com in October 2025. At the time it was read-only: agents could query your site but couldn’t touch it.
A second update in January 2026 added OAuth 2.1 authentication, making it easier to securely connect AI clients. In February, Automattic released an official Claude Connector, again read-only at the time. The current writing capabilities are the step toward which the platform has been building.
The feature is designed around explicit human approval. Before creating, updating, or deleting something, the agent describes exactly what it plans to do and asks for confirmation. New posts default to draft status, giving users the opportunity to review them before they are published; Modifying a published post triggers a warning that the changes will be visible immediately.
Deleting posts, pages, comments, and media sends items to the trash, where they can be recovered for 30 days. Categories and tags, which WordPress cannot delete, trigger an additional confirmation warning that the deletion is permanent. Each action is recorded in the site’s Activity Log.
User role permissions apply in full: an editor can create and edit posts, but cannot change site settings; a collaborator can write but not publish.
One of the most technically interesting aspects of implementation is knowledge of the subject. Before creating a page or post, the agent can read the site’s design system, colors, fonts, spacing, block patterns, and generate content that inherits those specifications.
Writing capabilities are available today on all WordPress.com paid plans. Users enable them through the MCP dashboard at wordpress.com/me/mcp, toggling the specific operations they want to allow on each site.
Supported clients include Claude, ChatGPT, Cursor and any other MCP-enabled tools. WordPress.com powers a significant part of the web; According to figures presented at Automattic’s State of the Word event in December 2025, WordPress runs more than 43% of all websites globally and has a 60.5% share of the content management systems market.
The scale at which writable AI agents can now operate on that infrastructure is considerable.
The MCP ecosystem has been expanding rapidly. The WordPress MCP Adapter, which enables similar functionality on self-hosted WordPress installations, has been moving towards inclusion in WordPress Core.
Automattic’s other products, including WooCommerce and Beeper, have their own MCP implementations. The pattern—standardized access by AI agents to application functionality, rather than one-off integrations—is becoming more of an architectural assumption than an experiment.
For WordPress.com users, the practical issue is trust. Giving an AI agent write access to a production site is a different proposition than asking it to summarize your traffic. Automattic has leaned into this explicitly, making the approval model the centerpiece of the ad and the granular operation toggles the default settings.
