What Is MCP
What the Model Context Protocol is, why it matters for connecting AI clients to WordPress, and how Protuno implements it through the WordPress Abilities API.
MCP (Model Context Protocol) is an open standard that lets AI clients, Claude, Cursor, VS Code Copilot, and others, connect to external tools and services through a defined interface. When an AI client is connected to an MCP server, it can call tools on that server to read data, take actions, and produce results.
For WordPress, this means an AI agent can interact with your site's content, pages, and design system directly, without you copy-pasting content between the AI and WordPress, and without the AI making guesses about your site's current state.
Why This Matters For WordPress
Without MCP, working with an AI on a WordPress site looks like this:
- You copy page content out of WordPress
- You paste it into the AI chat
- The AI responds with suggested changes
- You manually apply those changes back in WordPress
With MCP, the AI is directly connected to your site:
- The AI reads your page content, post data, and design directly from WordPress
- The AI makes changes (creates a page, updates a section, edits CSS) without you doing the copy-paste
- You see the result live on the site
How Protuno Implements MCP
Protuno does not run a standalone MCP server. Instead, it registers its capabilities through the WordPress Abilities API (available in WordPress 7.0+) and exposes them through a gateway that MCP-compatible AI clients can connect to.
Each capability Protuno registers is called an ability. Abilities are actions the AI can invoke, things like "create a page," "add a Proton section," "update a widget's code," "read the global CSS." Protuno registers 22 abilities under the protuno/ namespace.
When your AI client connects via MCP and the user asks it to "add a hero section to the home page," the AI calls the appropriate Protuno abilities in sequence, it does not need you to specify which API endpoint to use or what payload to send.
The 22 Protuno MCP Abilities
These are the actions your connected AI client can take on your WordPress site:
| Ability | What it does |
|---|---|
protuno/convert | Convert a Figma design or screenshot to a Proton section |
protuno/check-config | Check the current Protuno configuration and active features |
protuno/create-page | Create a new WordPress page |
protuno/add-section | Add a new Proton widget section to an existing page |
protuno/create-header-footer | Create a header or footer template in the Theme Builder |
protuno/update-code | Update the HTML, CSS, or JavaScript inside an existing Proton widget |
protuno/grab | Read the current code of a Proton widget |
protuno/set-globals-css | Update the Global Style CSS |
protuno/get-globals-css | Read the current Global Style CSS |
protuno/list-pages | List pages on the site |
protuno/get-page | Read the content of a specific page |
protuno/list-posts | List posts |
protuno/get-post | Read a specific post |
protuno/list-templates | List Theme Builder templates |
protuno/get-template | Read a specific template |
| And 7 more | For managing media, taxonomies, and site metadata |
This means a connected AI can traverse your entire site's structure, read any page, and make targeted edits to specific widgets, all through natural language instructions.
MCP vs The In-Editor AI Chat
Both the in-editor AI chat and MCP use an AI to work on your site. The difference is scope:
| In-editor AI chat | MCP (external AI client) | |
|---|---|---|
| Works on | One widget at a time, inside Elementor | Entire site, pages, templates, widgets, global styles |
| Where you work | Inside the Elementor editor | Your AI client (Claude Desktop, Cursor, etc.) |
| Scope of changes | Current widget's HTML/CSS/JS | Any page, any widget, any section |
| Available to | Free + Pro users | Pro users only |
| Requires | AI provider in Protuno Settings | MCP-compatible AI client + Protuno Pro |
Use the in-editor chat for focused editing of a single section. Use MCP when you want to build complete pages, run a full Figma-to-site pipeline, or manage content at scale without leaving your AI tool.
Frequently Asked Questions
Do I need to know anything about MCP to use it? No. You connect your AI client once (see Connect Any AI Client Over MCP) and then just talk to your AI in natural language. The protocol runs transparently in the background.
Is MCP specific to Protuno? No. MCP is an open standard. Many tools implement it, databases, APIs, design tools, file systems. Protuno adds WordPress and Protuno-specific abilities to an AI client that already knows the MCP protocol.
What AI clients support MCP? Claude Desktop, Cursor, VS Code (with Copilot), and a growing list of other tools. Any client that supports MCP can connect to Protuno's abilities.
Is this secure? Protuno's MCP abilities run through authenticated API endpoints. The AI can only perform actions that your Protuno user account has permission to perform. The connection uses your site's standard WordPress authentication, not a separate credential.