Get the plugin
WP MCP/Integrations/Block suites

Block suites integration

Otter Blocks MCP

Otter Blocks (themeisle-blocks/*) writes each post's stylesheet only when its editor calls back after a save, and serves it from post meta. WP MCP Pro's block suite pair reads Otter's block schemas, inserts and updates its blocks with every attribute checked and a correct unique id, and drops Otter's stylesheet meta and stale file after every write and rollback, so Otter rebuilds them on the next view. list-patterns browses the registered patterns Otter adds and import-pattern inserts one in a single undoable write.

block-suites-readblock-suites-writelist-block-typesadd-blockupdate-block

In practice

How Otter Blocks work goes with an agent.

Otter Blocks (themeisle-blocks/*) identifies each block with an id attribute shaped like wp-block-themeisle-blocks-name-8hex, and its stylesheet is written by the editor, not at render time. The editor calls Otter's REST route after a save, and Otter serves the result from post meta and a file in uploads. That is why an agent that edits block markup directly leaves stale CSS behind. WP MCP checks every attribute against Otter's registered schema, generates a correct id, and clears the stored stylesheet and its file after each write and rollback, so Otter rebuilds them on the next view.

Which Otter blocks can I use, and what exactly does one of them accept?

Runs block-suites-read get-block-schemas for the otter-blocks suite, once without a name for the list with attribute names, then with a block name for full definitions (type, default, enum). The result states that unknown attributes are refused.

Add this Otter block to the FAQ page. I do not have an id for it.

Runs parse-blocks for the content_hash, then block-suites-write insert-block. Attributes are checked first, a unique id is generated in Otter's shape, the post is snapshotted, and Otter's stylesheet meta and file are dropped.

Change the colour on the Otter block at path [1] but keep everything else.

Runs block-suites-write update-block with attrs. They are merged into the block's attributes and null removes one. The block's id cannot change, so an attempt is refused with unique_id_immutable.

Put an Otter pattern with remote images on /plans.

Runs list-patterns filtered to the otter-blocks suite, then import-pattern. Remote images are sideloaded into the Media Library first, the post is snapshotted once, and the pattern's blocks get fresh unique ids.

What your AI can do

Real Otter Blocks operations, every write reversible.

  • Read Otter block schemas
  • Insert and update Otter blocks with every attribute validated first
  • Rebuild Otter's per-post stylesheet after writes and rollbacks
  • Browse Otter's registered patterns and import one as a single undoable write
  • Roll back any block write exactly

Example prompts

Example Otter Blocks prompts.

List the Otter blocks and patterns available on this site.

Add an Otter accordion with these five questions to the FAQ page.

Import an Otter pricing pattern under the intro on /plans.

Under the hood

The Otter Blocks calls, under the hood.

wpmcp, otter-blocks

block-suites-read({ operation: "list-patterns", args: { suite: "otter-blocks" } }), patterns that use Otter

block-suites-write({ operation: "import-pattern", args: { name: "<pattern name>", id: 14, expected_hash, path: [2] } }), one snapshot, images sideloaded first

Safety and access

What can be undone on Otter Blocks, and what costs what.

Recoverable on Otter Blocks

  • insert-block, update-block and import-pattern each take one snapshot of the whole post, and rollback-operation restores the content exactly.
  • Otter's stylesheet meta and its stale CSS files are dropped again after a rollback, so Otter rebuilds them from the restored blocks.
  • Per-session rollback reverts all block writes of an agent run together.

Otter Blocks limits and opt-ins

  • Images sideloaded for an imported pattern stay in the Media Library after a rollback. Only the post content is restored.
  • import-pattern does not refuse attributes, because a suite's own pattern may carry attributes its current schema dropped. insert-block and update-block do refuse them.
  • A stale expected_hash is refused, so the agent must read the post again before writing.
  • The block suite pair registers only while a supported block suite is active.

Free or Pro: The Otter support is Pro: block-suites-read and block-suites-write register only with a Pro licence, and only while a supported suite is loaded. Free keeps the core Gutenberg tools, which edit any registered block by raw markup but do not check attributes against Otter's schema, generate its ids or clear its stylesheet. If you need schema checks, correct ids and an Otter-aware rebuild after every write and rollback, that is the Pro pair.

Part of WP MCP Pro

The Otter Blocks integration is part of WP MCP Pro.

One Pro licence unlocks the Otter Blocks 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

Otter Blocks MCP: questions, answered.

Is there a Otter Blocks free path?

Yes. The free Gutenberg block tools edit any registered block by raw markup. The dedicated Otter support, with validation, stylesheet rebuilds and pattern import, is part of WP MCP Pro.

Why is Otter's stylesheet deleted after a write?

Otter writes a per-post stylesheet only when its editor calls back after a save, and serves it from post meta. An agent write never triggers that call, so the old file would keep serving stale CSS. Deleting the meta and the files makes Otter rebuild it from the new blocks.

What does an Otter block id look like, and can the agent choose it?

It is wp-block-, then the block name with its slash turned into a hyphen, then eight characters from a UUID. The agent can leave it out or use the __UNIQUE_ID__ placeholder and the pack fills it in. Once a block has an id, update-block cannot change it.

With Otter Blocks, can the agent import patterns that are not Otter's?

Yes. list-patterns browses registered core, theme and suite patterns, and source directory browses the WordPress.org Pattern Directory, which has core blocks only and is cached for an hour. import-pattern inserts any of them in one snapshot with images sideloaded first.

What if the post changed since the agent read it?

Every write needs the expected_hash from parse-blocks. If the post changed since that read, the write is refused and nothing is saved, so an agent cannot overwrite an editor's newer changes. It reads the post again and retries.

Point your AI at Otter Blocks.