Get the plugin
WP MCP/Integrations/Page builders

Page builders integration

Oxygen (6 and classic 4.x) AI agent tool reference

Oxygen comes in two storage formats, and WP MCP handles both. Oxygen 6 runs on the Breakdance engine and keeps its JSON tree in _oxygen_data: WP MCP reads it and writes one node operation or a whole tree, byte for byte outside the edit, and refreshes the generated CSS through Oxygen when it is active. Classic Oxygen (4.x and earlier) stores a page twice, as a JSON tree and as a signed copy in ct_* shortcodes. A classic write keeps the two in step the way Oxygen's own save does: it adds an Oxygen revision, has Oxygen generate and sign the shortcode copy from the new tree, then refreshes the CSS. Because that signature uses the site's key, a classic write needs Oxygen loaded, and a page still in the pre-4.0 format, where the signed shortcodes are the only store, is read-only until it is saved in Oxygen 4. Every row is post meta, so rollback restores tree, shortcode copy and revisions together.

detect-builderget-builder-contentupdate-builder-content

In practice

How Oxygen (6 and classic 4.x) work goes with an agent.

Two builders share the Oxygen name. Oxygen 6 runs on the Breakdance engine and keeps its tree in _oxygen_data, with its own post types for templates, headers, footers, Components and design library parts. Classic Oxygen (4.x and earlier) keeps a JSON tree in ct_builder_json, moved to _ct_ keys from 4.8.3, plus a signed shortcode copy. WP MCP routes by the format it detects: Oxygen 6 gets node ids, a hash lock, revisions and Components, while classic gets path addressing and Oxygen's own serializer for the signed copy. A page from before 4.0 is read-only.

Save the hero as a Component and reuse it on the offers page.

Runs update-builder-content with operation save_template, the hero's node id and element {title}, which creates a published oxygen_block post and leaves the page alone. Then operation add with element {component: id} places one instance. The save is recorded as a creation, so rollback-operation trashes the Component.

Insert the Pricing part from the design library below the intro.

Runs get-builder-content with scope templates to find the part's id, then update-builder-content with operation add and element {template: id}. The template's elements are copied in with new node ids, and the write is pinned to the page's hash and snapshotted first.

Roll the landing page back to last Tuesday's revision.

Runs get-builder-content, whose read lists up to 20 revisions that carry the Oxygen data row, then update-builder-content with operation restore and path set to the revision id. The write needs the current hash and is a normal snapshotted operation.

Change the heading on the classic Oxygen homepage.

Runs detect-builder, which reports oxygen-classic, then update-builder-content with operation update, the element's path and text. Oxygen adds a revision, generates and signs the shortcode copy itself, and the CSS is refreshed. Dynamic data shortcodes in new text are signed through Oxygen too.

What your AI can do

Real Oxygen (6 and classic 4.x) operations, every write reversible.

  • Detect Oxygen 6 and classic Oxygen pages
  • Read a page's element tree in either format
  • Update, add, remove or move one element, or replace the whole tree
  • Classic: keep the tree and Oxygen's signed shortcode copy consistent, with an Oxygen revision
  • Refresh Oxygen's generated CSS after writes and rollbacks
  • Roll back any layout write exactly

Example prompts

Example Oxygen prompts.

Which pages are built with Oxygen, and which are classic Oxygen?

Update the heading text in the Oxygen hero section on the homepage.

Remove the old promo banner element from /offers.

Under the hood

The Oxygen calls, under the hood.

wpmcp, oxygen

detect-builder({ post_id: 63 }), reports oxygen or oxygen-classic

update-builder-content({ post_id: 63, builder: "oxygen-classic", operation: "update", path: "0.2", text: "New heading" }), needs Oxygen loaded; snapshotted

Safety and access

What can be undone on Oxygen, and what costs what.

Recoverable on Oxygen

  • Oxygen 6 data and its generated CSS rows are post meta, so a post snapshot restores them and the rollback rebuilds the page CSS.
  • Classic pages restore the tree, the signed shortcode copy and the Oxygen revision rows together, as the consistent set they were.
  • Saving a Component is recorded as a creation, and rollback-operation moves the new post to the trash.

Oxygen limits and opt-ins

  • Oxygen 6 writes need Oxygen 6 active and expected_hash from the latest read on any page that has a layout.
  • Oxygen 6 design data (classes, variables, palette) is read-only through this tool.
  • Classic writes need Oxygen loaded when the page has a signed shortcode copy, and a page still in the pre-4.0 format is read-only.
  • If a loaded Oxygen is not version 6, the read carries a version note asking the agent to check edits in the editor.

Free or Pro: Oxygen 6 and classic detection, reads, writes, revisions, Components and the template scopes are all Pro, with no opt-in filter. On the free tier, export-content with mirror:true writes Oxygen pages to JSON files for git: classic pages as their JSON tree, with pre-4.0 shortcode-only pages written as lines that a restore refuses, since only Oxygen can sign them.

Part of WP MCP Pro

The Oxygen integration is part of WP MCP Pro.

One Pro licence unlocks the Oxygen tools and every other Pro integration on this site: Elementor deep editing, the other page builders, the forms adapter pack and more, 133 Pro tools in all. The free plugin still covers the rest of WordPress and the full snapshot and undo engine, so you can try the safety model before you buy.

Frequently asked

Oxygen MCP: questions, answered.

Why do classic Oxygen writes need Oxygen active?

Classic Oxygen signs its shortcode copy with the site's key, so only Oxygen's own serializer can generate a valid one. WP MCP never forges that signature; it asks Oxygen to produce it.

What about pages from before Oxygen 4.0?

Those hold only the signed shortcodes, with no JSON tree, so they are read-only here. Open and save the page in Oxygen 4 and it becomes editable.

How does it tell Oxygen 6 from classic Oxygen?

By what the page stores. A valid tree in _oxygen_data means Oxygen 6, reported as oxygen. A JSON tree or shortcode copy under the ct_builder keys, with or without the _ct_ prefix, means classic, reported as oxygen-classic.

Can it edit Oxygen 6 headers, footers and templates?

It can read them with the templates scope (template, header, footer, block and part posts) and copy their elements into a page. Its writes target pages and new Components, not the template posts themselves.

Does the Oxygen 4.8.3 meta key change matter?

No. From 4.8.3 the rows moved to _ct_ keys, and WP MCP uses whichever layout the loaded Oxygen reports or the page already has, so older and newer classic sites both work.

Which Oxygen 6 version is it built for?

The adapter was recorded from the Oxygen 6.1 source. Each Oxygen 6 read reports the loaded version, and if the major version is not 6 it adds a note to check edits in the Oxygen editor.

Is there an AI agent for Oxygen?

Yes, through WP MCP rather than Oxygen itself. An agent you choose, such as Claude or Cursor, can detect Oxygen 6 and classic Oxygen pages. WP MCP is a separate plugin, not a built-in AI feature of Oxygen. On the undo side: Oxygen 6 data and its generated CSS rows are post meta, so a post snapshot restores them and the rollback rebuilds the page CSS.

Point your AI at Oxygen (6 and classic 4.x).