The Divi AI Agent ships with more than 190 purpose-built tools for Divi and WordPress - far too many to memorize, and too many to reasonably list one by one in an article.
This is the map instead: the broad categories of work the Agent can do, how to browse its full tool catalog when you want to see the actual list, and what its risk badges mean before you approve anything. If you're after ready-made requests to try, this collection also has a prompt library.
Browsing the Tool Catalog
Every tool the Agent can use is listed in the Tools catalog 
.
Open an AI Chat window by clicking on the AI icon
in the left sidebar. Click the Tools icon
to open the tools catalog.
Inside, you'll find:
A search box (placeholder text "Search tools...") that filters the list as you type, matching against each tool's name and description.
A risk filter that narrows the list to one risk level at a time.
Two groups:
Agent (the tools available in Build mode's full planning-and-execution flow)
and Ask (the smaller, read-only set available in Ask mode, limited to searching and reading Divi's documentation). Each group shows how many tools it currently contains.
Each entry in the list shows the tool's name, a plain-language description of what it does, and its risk badge.
What the Risk Badges Mean
Every tool in the Tools catalog carries one of three risk badges. Hover a badge to see its tooltip, quoted here exactly as it appears:
Badge | Tooltip |
Safe | Safe actions do not require confirmation |
Moderate | Moderate actions require confirmation when configured |
Destructive | Destructive actions require confirmation before execution |
Destructive tools always pause and ask you to approve them before they run.
Moderate only pauses when the individual tool's own definition calls for it, and which Moderate tools do that isn't something you can turn on or off anywhere in Divi - it's built into the tool, not a setting you control. Check this [Name of article] to learn more about the approve mode.
Adding a Tool as Context
If you want the Agent to use a specific approach rather than deciding for itself, you can pin an individual tool into the conversation.
Open the Tools catalog.
Find the tools you want to use.
Click on the @ symbol to add the selected rule to the context window.
What the Agent Can Do
The Agent's tools fall into about a dozen broad categories. None of these lists are exhaustive - each is a sample of what the category covers.
Page structure and modules
The largest category. The Agent can read your page's full hierarchy of sections, rows, columns, and modules, then add, edit, move, clone, or remove any of them.
Get the complete page hierarchy of sections, rows, columns, and modules
Clone an existing module, or lock one so it can't be edited by accident
Search modules by their content, or find repeated styles worth turning into a preset
Add interactions between modules - for example, having one module's hover or click affect another
Presets
The Agent can work with Presets the same way you would by hand.
Create a preset from a module's current styles
Save pending preset changes
Clone or delete a preset
Assign a preset as the default applied to new modules of that type
Global variables and colors
The Agent can read and manage Global Variables, including the global color palette.
Create, update, or delete a global variable
Set a global variable's value everywhere it's used at once
Add one or more global colors to your palette
Layout generation
Rather than adding elements one at a time, the Agent can generate a whole structure from a description.
Generate a page, section, row, or column layout from a plain-language brief
Apply a premade Flex or CSS Grid structure template to existing elements
Images and media
The Agent can generate, find, and evaluate images, building on the same concepts covered in Aspect Ratio, Image Framing, and Image Presets.
Generate custom AI imagery and import it straight into your Media Library
Fetch stock images from Unsplash
Analyze what's actually in an image you've attached
Compare your current design against a reference image
Search your existing Media Library for something you've already uploaded
Loops and dynamic content
The Agent can set up Loop Builder and Dynamic Content the way you would by hand.
Enable or disable a loop on a module
Configure a loop's content source and result order, and set filters on it
Build Dynamic Content tokens from a variable definition
List the Dynamic Content options available for the current post
Canvases, workspaces and the builder interface
The Agent can drive the builder interface itself, not just the content on your page - including Canvases and Workspaces.
Create, duplicate, or delete a canvas
Switch or duplicate a workspace
Change the canvas zoom level, or switch breakpoint or Wireframe view
Set the interface to light or dark mode
Pages and posts
Beyond a single page's content, the Agent can work at the site level with pages, posts, and custom post types.
Create a WordPress page, post, or custom post type item
Search existing pages, posts, or custom post type items
Change a page or post's publication status
Update a page's content or settings directly
Menus
Create a WordPress navigation menu by name
Assign a menu to a theme location
Reorder items within a menu, or add and remove items
Theme Builder and Theme Options
The Agent can work with Theme Builder templates and Theme Options directly.
Create a Theme Builder template
Set a template's display conditions - which pages it applies to
Read your site-wide Theme Options
Update an individual Theme Option
Divi Library
Save a module, row, or section to the Divi Library
List your saved library items, filtered by type or name
Replace the current page with a saved library layout
Documentation and troubleshooting
Search Divi's product documentation for an answer
Retrieve the full text of a specific documentation article or knowledge article by name
Agent configuration
The Agent can manage its own standing configuration - the Rules and Commands covered in earlier articles in this collection - as well as your responsive breakpoints.
Create or update an AI rule or an AI command
Update your responsive breakpoint settings
How the Agent organizes its own work
A last category that isn't about your page at all - it's how the Agent manages a request behind the scenes.
Keep a running to-do list of the steps it has planned and completed for your request
Run up to ten independent sub-agent tasks at once and combine their results
Ask you a clarifying question when it's missing information it needs to continue
Request additional tools if the ones already available can't finish the job
List its previous build activity from earlier in the same chat
Where the Agent Stops
The Agent is broad, but it isn't unlimited. A few honest boundaries worth knowing before you rely on it:
Site-level requests are narrower than in-builder requests. Working from inside the Visual Builder gives the Agent its full range described above; requests made from outside the builder draw on a smaller set of the same abilities.
It builds toward a reference, not a pixel-perfect copy of one. When you hand it a reference image to match, it compares its work against that image and aims for a similar design using real Divi sections, rows, and modules - not an exact recreation down to the pixel.
It doesn't connect to outside tools through MCP. Support for the Model Context Protocol, which would let the Agent call external services beyond Divi and WordPress, hasn't shipped yet.