WordPress.com MCP Lets AI Agents Write and Publish Posts

WordPress.com MCP write capabilities let AI agents like Claude draft blog posts, manage pages, and update your site content through natural conversation. 

WordPress.com has expanded its Model Context Protocol (MCP) support with write capabilities that turn AI agents into active site collaborators. Where the original WordPress.com MCP integration introduced last October let tools like Claude or ChatGPT read site data, this update adds 19 new writing abilities across posts, pages, comments, categories, tags, and media. The shift changes how designers and content teams interact with their WordPress sites.

The new WordPress.com MCP write tools work through natural conversation. A designer can tell Claude to draft a post, categorize it, add tags, and write a meta description in one message. The agent handles the sequence without the user opening a dashboard. Posts default to drafts so nothing goes live without review, and deletions move to trash for 30 days rather than removing content permanently.

WordPress.com MCP AI agent write capabilities interface showing natural conversation to create content
WordPress.com MCP write capabilities let AI agents create posts, build pages, and manage site content through natural conversation.

Design-Aware Content Through WordPress.com MCP

One notable feature is design awareness. Before building a page, the WordPress.com MCP agent queries the active theme to understand colors, fonts, spacing, and block patterns. Generated content inherits those specs automatically. When the theme changes, outputs adapt without manual rework. That integration keeps AI-generated pages consistent with the visual system the designer already established.

WordPress.com MCP prompt example showing AI agent asked to build a landing page with theme-aware design specs
WordPress.com MCP reads the active theme before building pages, so AI-generated output inherits your site's visual language automatically.

WordPress.com MCP write access is available now on all paid plans. Users enable it through the MCP dashboard at wordpress.com/me/mcp and connect a compatible AI agent. No additional installation is required beyond what the original MCP integration needed. Compatible clients include Claude, Cursor, ChatGPT, and any other tool that supports the MCP standard.

WordPress.com MCP output showing a completed landing page built by an AI agent using theme-aware block patterns
A landing page built by an AI agent through WordPress.com MCP, using the site's existing block patterns and design system.

Practical Impact for Design Teams

For design teams managing content at scale, WordPress.com MCP offers a practical shortcut. Approving comments, fixing image alt text across a library, reorganizing taxonomies, or setting up new category structures becomes a conversation rather than a workflow. The approval step stays with the user, so the agent cannot publish or delete without confirmation.

The write capabilities follow months of feedback from users who found read-only MCP support useful but incomplete. The protocol is positioned to grow, with Automattic signaling more content types and operations on the roadmap. For teams already inside the WordPress ecosystem, the calculus is straightforward: 19 new operations in the same tool they already use, no new interface to learn.

Get Featured

Send your project to be featured on the blog. Follow the instruction on the template and good luck. Ah, make sure you add the images and credits that are due.

Submit content