Description
WordPress.com has a built-in MCP server. Now self-hosted WordPress does too.
IATO MCP connects your WordPress site to Claude Desktop and other MCP-enabled AI clients. Once connected, you can ask Claude to audit your site and fix SEO issues, identify orphan pages, clean up broken links, and more, all in a single conversation.
How it works
- Install and activate the plugin
- Follow the setup wizard — copy the config into Claude Desktop, or use “Add Custom Connector” with your site URL
- Connect your IATO account for AI-powered analysis (free trial up to 500 pages)
What Claude can do
Without an IATO account (64 WordPress tools):
- Read and edit posts, pages, and media
- Create new posts and pages with excerpt support
- Update SEO titles and meta descriptions (Yoast SEO, RankMath, SEOPress)
- Update canonical URLs
- Update image alt text
- Upload new images to the media library (base64 by default; URL ingestion optional with admin allowlist + full SSRF protection)
- Set and clear the featured image on any post
- Read and write arbitrary post meta with a credential-key denylist and a known-safe theme/builder/SEO allowlist (force=true required outside the allowlist)
- Set per-post theme + builder page settings (hide title, sidebar layout, content layout, etc.) on Astra, Kadence, GeneratePress, and Elementor in one call
- Give one page its own CSS, fonts and scripts, and make expiring preview links for drafts (switched off until an administrator turns them on; a script runs only after an administrator approves it)
- Read and edit navigation menus, create menus and assign them to theme locations
- Nest pages under a parent and change the permalink structure, with 301 redirects from the old URLs and one-step undo
- Manage categories, tags, and taxonomy terms
- Manage JSON-LD structured data
- Manage redirect rules
- Read and write Elementor page builder data
- Widget-grained Elementor edits with optimistic concurrency, idempotency, and bulk operations
- Fix heading levels, text, links and image alt text on Elementor Atomic Editor (V4) elements (Elementor 4.3.0 or later), converted and validated with Elementor’s own code, each change with its own rollback
- Clone the styling of an existing Elementor post in one call via
update_elementor_data(..., inherit_settings_from: <id>) - Create Elementor Pro Theme Builder templates and set their display conditions, such as a site-wide header and footer
- Resolve URLs to their rendering post, and to the Theme Builder header and footer they get
- Search content across the site
- Read site info and settings, and set the site title, tagline and home page
- Read and filter comments
- One-call rollback for any tracked write — every change emits a receipt with a stable
change_id; pass it back to therollbacktool and the original value is restored - Purge page caches after every change (LiteSpeed Cache, WP Rocket, W3 Total Cache, WP Super Cache, WP Fastest Cache, SiteGround Optimizer) and say so in the response, including any cache IATO cannot reach, such as a CDN
With an IATO account (12 bridge tools — full analyze-and-fix pipeline):
- Start a new crawl of your site directly from Claude (admin only)
- Check crawl status and list recent crawl jobs
- Run a full SEO audit and fix title, meta description, and alt text issues automatically
- Identify orphan pages not linked from any navigation menu
- Audit navigation menus for gaps and missing sections
- Surface thin content with specific improvement recommendations
- Map broken links to source posts for direct editing
- Analyze site taxonomy and suggest consolidations
- Get AI-prioritized suggestions across all areas
- Flag slow pages with contributing performance factors
Supported SEO plugins
- Yoast SEO
- RankMath
- SEOPress
- Falls back to native WordPress title if none detected
Example prompts
“Crawl my site and fix all missing meta descriptions”
“Show me pages that aren’t in any navigation menu and add them to the right place”
“What are the most impactful improvements I can make to my site right now?”
“Find all broken links and tell me which posts contain them”
“Audit my categories and tags and suggest consolidations”
“Set every H2 heading in these Elementor posts to H1”
“Find all button widgets on the site and change their color to #ff0000”
External Services
This plugin connects to the following external service when configured:
IATO API (https://iato.ai) — When you enter an IATO API key in the plugin settings, the plugin sends requests to https://iato.ai/api to retrieve crawl data, SEO audit results, sitemap information, and AI-generated improvement suggestions. No data is sent to IATO until you configure an API key. Your public page URLs (as crawled by IATO) and crawl analysis results are transmitted.
The plugin also implements an OAuth 2.0 authorization server on your WordPress site so that MCP clients like Claude Desktop can authenticate via the standard “Add Custom Connector” flow. This communication stays between the MCP client and your WordPress site — no data is sent to third parties during authentication.
Screenshots




Installation
- Upload the plugin files to
/wp-content/plugins/iato-mcp/or install via the WordPress plugin directory - Activate the plugin via the Plugins menu in WordPress
- Follow the setup wizard that appears — it provides the JSON config for Claude Desktop
- In Claude Desktop, either paste the JSON config or use “Add Custom Connector” and enter your site URL
- Optionally, go to Settings > IATO MCP to enter your IATO API key for the full analysis pipeline. Use a separate IATO API key for each client site: a key gives access to every crawl in its IATO account, so a key shared between sites lets one site’s administrators see the others’ crawl data.
For detailed setup instructions, see the IATO MCP documentation.
FAQ
-
Do I need an IATO account?
-
No. The plugin works standalone for reading and editing WordPress content with 57 built-in tools. An IATO account (free trial up to 500 pages) unlocks 12 additional bridge tools: start/list/status crawl management, SEO audit, broken links, content gaps, orphan pages, navigation audit, taxonomy analysis, AI suggestions, and performance reports. They are available to administrators. If you manage several client sites, give each site its own IATO API key: a key can read every crawl in its account.
-
Which WordPress version is required?
-
WordPress 6.2 or higher with PHP 8.0+. The plugin uses the WordPress REST API and implements OAuth 2.0 for secure authentication with AI clients.
-
Yes. The plugin uses standard HTTP requests (one per MCP call) rather than long-lived connections, so it works on all hosting environments including shared hosting.
-
Which AI clients are supported?
-
Any MCP-enabled client: Claude Desktop, Cursor, VS Code with GitHub Copilot, and any client that supports the Streamable HTTP MCP transport.
-
How does authentication work?
-
The plugin generates a secure API key on activation. You can authenticate in two ways: paste the provided Bearer token config into your AI client, or use Claude’s “Add custom connector” flow, which handles OAuth 2.1 with PKCE automatically. Each OAuth connection gets its own expiring token, acts as the administrator who approved it, and can be revoked on its own under Settings > IATO MCP > Connected OAuth clients.
-
Why does the plugin support two auth methods?
-
AI clients like Claude Desktop authenticate via a WordPress Application Password (or the OAuth 2.0 / PKCE flow), which is the WordPress-native pattern most users will use. The plugin also accepts the plugin-generated Bearer token at the same MCP endpoint — that path is used by the IATO platform’s own integrations (for example, the dashboard’s “Sync pages, posts, menus, and taxonomy from WordPress” feature, which composes the plugin’s read tools to pull content into IATO). Both methods land at
/wp-json/iato-mcp/v1/messageand are validated byclass-auth.php. You don’t have to choose — paste your Bearer token into the IATO platform connection, generate an Application Password for Claude Desktop, and the same plugin handles both. -
Is my content sent to IATO or Anthropic?
-
WordPress content (post titles, meta descriptions, etc.) is never sent to IATO. IATO crawls your public URLs the same way a search engine would. Claude processes content within your AI client session only. The IATO API is only called when you use bridge tools, and only crawl analysis data (not your content) is transmitted.
-
A page lost an embed or script after an IATO edit. How do I repair it?
-
Open the page in Elementor, go to History, then Revisions, and apply the last revision saved by a person before the first IATO edit. Then re-apply the IATO changes you want to keep. If revisions are disabled on the site, there is no earlier copy to restore.
-
A page edited with IATO shows a missing link or image, or Elementor will not save it
-
IATO 1.13.0 and earlier could store a plain value on an Elementor Atomic Editor (V4) element, such as a heading, paragraph, image or button built with the Atomic Editor, where Elementor expects a typed value. Elementor then drops a link or image it cannot read, and saving the page in Elementor fails with “Settings validation failed”.
To find the affected elements, read the page with get_elementor_data, get_elementor_widget, list_elementor_widgets or find_elementor_widgets. Since 1.13.1 they flag the element with stored_value_invalid: true, and invalid_keys names the settings to repair; get_elementor_data also lists them under invalid_elements.
To repair an element, send each listed setting again with update_elementor_widget, set_widget_setting or set_heading_level. From 1.14.0 (with Elementor 4.3.0 or later) the plain value works, for example {“tag”: “h2”} or {“link”: {“destination”: “https://example.com/”}}: IATO stores it in Elementor’s format and the element is repaired. With 1.13.1, or on a site without Elementor 4.3.0, send it as a typed value, in the shape get_elementor_widget shows under settings for the element, for example {“tag”: {“$$type”: “string”, “value”: “h2”}}. When several elements on a page are affected, repair them one at a time; Elementor saves the page normally again once the last one is repaired. A damaged element cannot be copied (update_elementor_patch refuses the copy): repair the source first, then copy it. You can also restore a revision saved before the IATO edit, under History, then Revisions, in Elementor.
-
Can IATO edit Elementor Atomic Editor (V4) elements?
-
Yes, from 1.14.0, with Elementor 4.3.0 or later and the Atomic Editor turned on. set_heading_level, set_widget_setting, update_elementor_widget, update_elementor_widgets_bulk and update_elementor_patch take plain values for atomic headings, paragraphs, buttons, images and containers, and update_alt_text takes post_id and element_id for an atomic image. IATO converts and validates each value with Elementor’s own code and keeps every setting the call does not name as it was stored. Changes go live, like edits to classic widgets, and each one can be rolled back on its own.
The alt text of an image from the Media Library belongs to the attachment, so IATO changes it there, and the change applies everywhere that image is used. Global classes and component instances are not edited by IATO yet. On sites with an older Elementor, IATO accepts only typed values for atomic elements and says what is missing. While the Atomic Editor is off, IATO does not change atomic elements at all, because Elementor cannot check them; the rest of the page can still be edited.
-
Why do visitors still see the old page after a change?
-
IATO purges the cache plugins it finds after each change and says what happened in the tool’s
cacheobject. A cache in front of WordPress, such as Cloudflare or a host’s CDN, is out of its reach: the response names it incache.notesso the AI client can tell you to purge it from that service’s dashboard. On Hostinger, LiteSpeed’s full purge also clears hCDN when Hostinger Tools 3.0.79 or later is active. A single-page purge cleared it too on the Hostinger site we tested, but IATO does not control that, so if a page still looks stale, purge hCDN in hPanel. -
Why was a style block or an SVG removed from my HTML widget?
-
Content written through MCP is filtered, whatever the role of the user behind the connection, and
<style>and<svg>are not on the list of allowed tags. The response’ssanitized_reportsays what was removed from each value. To keep them in Elementor HTML widgets, turn on “Allow inline SVG and style blocks in HTML widgets for administrators” under Settings > IATO MCP. It applies only to a WordPress user who holds bothunfiltered_htmlandmanage_options, and needs WordPress 6.5 or later. Scripts, event attributes,javascript:URLs,foreignObject,@importand off-siteurl()are still removed. -
Can I control which tools are available?
-
Yes. Go to Settings > IATO MCP to enable or disable individual tools. You can turn off any tool you don’t want AI clients to access.
-
Can AI clients upload arbitrary files to my media library?
-
Only images, and only when the calling user has the
upload_filescapability. Thecreate_mediatool enforces an image-only MIME allowlist (JPEG, PNG, GIF, WebP, AVIF) verified against actual file bytes — the claimed mime_type is never trusted. SVG uploads are not supported in this release. Files exceeding the size cap (default 10MB) or the dimension cap (default 8000×8000) are rejected, as are filenames containing.php,.phtml, or.htaccess. URL-source ingestion is disabled by default; admins who enable it must also configure a host allowlist, and private/loopback/cloud-metadata IPs are rejected even for allowlisted hosts. Each upload counts against a per-user rate limit (default 20/min) and emits achange_receipt— rolling back fully deletes the attachment file. All four limits are configurable from Settings > IATO MCP.
Reviews
There are no reviews for this plugin.
Contributors & Developers
“IATO MCP” is open source software. The following people have contributed to this plugin.
ContributorsTranslate “IATO MCP” into your language.
Interested in development?
Browse the code, check out the SVN repository, or subscribe to the development log by RSS.
Changelog
1.21.1
- Fix:
update_elementor_patch:addat a list index now inserts before the item there, as RFC 6902 says and the tool’s description promised. Since 1.3.0 it replaced the item, so a widget added at an index deleted the widget that stood there.copyto an index overwrote the item there andmoveto an index deleted one; both now insert. An index past the end of a list is refused instead of leaving a gap, and an element cannot be moved into its own children. Usereplaceto overwrite. A page that lost an element this way can be put back by rolling back that write’s receipt within 90 days, or from Elementor’s revisions. - Fix:
update_elementor_patch:removeof an item in the middle of a list left a gap in its numbering, so a later op in the same patch could not find the next item, and the list could be stored as a JSON object. - New: each
update_elementor_patchop on a list item returnslist(where it acted, the list’s length before and after, and its items’ ids afterwards), on dry runs too, so a misplaced insert shows before it is written. - Fix:
stored_matches_inputno longer reads a different key order as a different value. Elementor’s save writes an element’s keys in its own order, so a real write said false where its dry run said true. - Fix: an
addto/-, and an op whose position a later op in the same patch shifted, now getstored_matches_input. They used to get nothing, or be compared with whatever stood at the old position. - New: when the stored value differs,
stored_differencesays how:markup_normalized(the same HTML written differently, such as single quotes rewritten as double quotes),keys_addedorkeys_removed(withadded_pathsandremoved_paths; Elementor’s save dropsisInnerfrom a widget), orchanged. Asanitized_reportentry for a value WordPress rewrote without removing anything now says so; it used to be empty. - Fix:
get_page_fontsandset_page_fontsreport where each Google font is served from (served_from:local,local_after_first_vieworgoogle). Elementor serves a local copy it already holds even with “Load Google Fonts Locally” off, sogoogle_fonts_localwas false on sites that served fonts locally. It now means every Google font on the page comes from this site;local_google_fonts_settinggives the setting itself. - New: Settings > IATO MCP > Tools says that AI clients read the tool list when they connect, and saving a change to the tools says to reconnect the connector. This server cannot tell a connected client that the list changed.
1.21.0
- New:
set_page_cssandget_page_cssgive one page its own CSS, printed in that page’s head only. They follow the same rules as Additional CSS: nothing loaded from another site, at most 200 KB, and anything that would close the style element removed. The CSS is checked again when it is printed. Needs a WordPress user who may edit CSS (administrators and editors) andedit_poston the page. - New:
set_page_fontsandget_page_fontsload Elementor font families on one page only, through Elementor’s own font loader, so “Load Google Fonts Locally” is respected. Families Elementor does not know are refused. - New:
create_preview_linkmakes an expiring URL (15 minutes by default, an hour at most) that shows a draft, pending or scheduled post without a login, so an AI client can check its work. Responses to it are not cached and carry noindex and no-referrer. Rolling back its receipt revokes it, and publishing or trashing the post ends it. Needs permission to publish the post. - New:
set_page_scriptandget_page_scriptstore inline JavaScript for one page. It runs only after an administrator reads it and approves it in wp-admin with their password, and any change waits for approval again. It needs the new “Allow page scripts” setting, off by default, and an administrator who may post unfiltered HTML.<script,</script,<!--and invisible or direction-changing characters are refused, and the approval screen marks every character outside plain ASCII. Turning the setting or the tool off, or deactivating the plugin, stops every page script and purges those pages, even with cache purging off. - New: These seven tools ship switched off. An administrator turns them on in Settings > IATO MCP > Tools. A new “Page assets” tab lists every page with page CSS or fonts and every live preview link, with buttons to remove or revoke them. The Plugins screen warns before the plugin is deactivated, since those pages lose their CSS and fonts while it is off. Uninstalling keeps page CSS and fonts.
- New: HTML in Elementor widgets keeps every
aria-*attribute,inert,disabledandreadonlywhere HTML has them, and<input>with ordinary types (not password, file, image, hidden or submit). Nothing that passed before is removed now: a live check runs a fixture set through 1.20.0’s filter and this one. - New: With “Allow inline SVG and style blocks” on, HTML widgets also keep
ellipse,text,tspan,defs, gradients,stop,marker,clipPath, andusepointing at#idin the same SVG. A@propertysyntaxdescriptor such assyntax:'<number>'is kept in style blocks outside an<svg>(WordPress 6.7 or later). - New: Elementor write responses say for each change whether the stored value is what was sent (
stored_matches_input). When it is not, they say what was stored.update_elementor_widget,update_elementor_patchandupdate_elementor_datatakereturn: "minimal", which returns sizes and hashes instead of whole values. - Fix:
sanitized_reportnow names style declarations WordPress removed fromstyleattributes. When Elementor’s own filter removed more for a user withoutunfiltered_html, that used to be left out of the report. - Fix: a removed
<style>block’s report namesset_page_cssand where to turn it on. - Fix: With “Allow inline SVG and style blocks” on, a value with markup inside an SVG
<title>is no longer kept by the opt-in: a browser reads that markup, while the filter read it as text. Such a value now goes through the ordinary rule.
1.20.0
- New:
update_postandcreate_posttakeparent, for pages and other hierarchical types, andmenu_order. A page’s URL becomes its parent’s URL plus its slug, so /services/web-development/ can be built.parentis refused for posts (parent_not_supported). A page cannot go under itself or a page below it (parent_cycle). The parent must be a page of the same type that the caller can read. A parent that is not published is a warning. - New: A new parent moves the page and every page under it. When any of those URLs can be opened now, the call needs
confirm_url_break. The dry run lists every URL that moves (url_changes) and the redirects it would create. By default each published page’s old URL gets a 301 to its new one (redirect_old_urls), through the same redirect pluginupdate_redirectuses. Creating those redirects needs an administrator; an Editor can passredirect_old_urls: false. One receipt covers the parent, the order and a slug changed in the same call, and its rollback removes the redirects before putting the page back. The whole site’s cache is purged. - New:
update_site_settingstakespermalink_structure,category_baseandtag_base, andget_site_settingsreturns the two bases. A change moves every post and the category, tag and author archives, so it needsconfirm_url_break. The dry run reportsmoved_countand the first 25 URLs. 301 redirects from the old URLs are created by default, up to 1,000; with more, or more than the built-in list or Safe Redirect Manager can hold, none are created and a warning says so. Plain permalinks are refused, and so is a first change away from plain or /index.php/ permalinks, which belongs in Settings > Permalinks. Rollback restores the settings and removes the redirects. - New: The rewrite rules are rebuilt by the next request after a permalink change, as wp-admin does, not in the request that made it. Rebuilding them there would store rules without the ones other plugins add, such as WooCommerce’s account pages, and with category archives at the old URL.
- New: Moving a page back, or changing the permalinks back, with a new call rather than a rollback takes away the redirects the first change left, which would otherwise loop. A redirect someone else set on a URL that a change would make live is refused with
new_url_redirected. - New: A permalink change whose new post URL is already a published page’s URL is a warning, and so is a post slug that does the same: WordPress shows the page there. A page at the start of the new structure, such as /blog/ for /blog/%postname%/, keeps working, including /blog/page/2/.
- New:
get_postandget_postsreturnparentandmenu_order. - Fix:
update_postrefused a page’s slug when any post of any type used it, such as a menu item called “services”. It now uses WordPress’s own rule: a page is compared with pages and attachments under the same parent, a post with posts. A draft is checked as if published, so the slug is not renamed “-2” when it goes live. A slug WordPress reserves, such asfeed, is refused with the slug WordPress would use instead. - Changed: Server instructions add a “Restructuring URLs” section (parent pages first, dry run, apply, then update links in content and menus), and on Elementor sites list
slug,parentandmenu_orderamong the fieldsupdate_postmay change.
1.19.3
- Fix: On sites with Elementor, the instructions the plugin gives AI clients said
update_postcould change onlystatus,titleandexcerpt, so clients were told not to change a post’s date or author with it. The list now includesdate,date_gmt,allow_futureandauthor, and says content edits go through the Elementor tools. - Tests: a unit test checks that every
update_postfield the instructions name is in its input schema.
1.19.2
- New:
get_site_inforeturnsplugin_version, the IATO MCP version running on the site, so one call confirms an update is live. Every caller sees it.
1.19.1
- Changed: The tool descriptions clients read are clearer about publish dates and authors.
create_post‘s description mentions them,dateanddate_gmtsay they are ISO 8601 and thatdate_gmtis worked out fromdatewhen omitted, andallow_futuresays what is refused and when a post is scheduled. - Changed:
create_media‘s description sends files over about 3 KB tomedia_upload_start,media_upload_chunkandmedia_upload_finishinstead of URL ingestion, and says the response carriessha256anddecoded_bytes. - Fix:
get_iato_sitemapreadscrawl_idbut did not list it, so clients could not see it. It is listed now. The WordPress ability already listed it. - Fix:
create_poststores no excerpt, but anexcerptsent with it could appear insanitized_reportas text that was removed. It is now ignored, as the schema says. - Tests: a unit test checks that every tool’s input schema lists exactly the arguments its handler reads, so the two cannot drift apart.
1.19.0
- New:
update_postandcreate_posttakedate(the site’s local time),date_gmt(UTC),allow_futureandauthor. A published post given a future date is refused withfuture_date_on_publishedunlessallow_futureis true, when it is scheduled and the response says so. A draft keeps its status and uses the date when published. Rollback restores the date, the UTC date and the status together. Changing an author needsedit_others_posts, and the new author must be able to edit that type of post. - New:
get_postsreturnsdateanddate_gmt, andget_postreturnsdate_gmt. - Changed:
get_post‘sdateis now the site’s local time, so that it can be sent back toupdate_post. It was UTC: readdate_gmtfor that.modifiedis still UTC. - Changed: Publishing a draft that has a future date is refused with
future_date_on_publishedunlessallow_futureis true. WordPress scheduled it without saying so. - New:
create_mediatakessha256, and refuses a file that arrives altered withchecksum_mismatch. It always returnssha256anddecoded_bytes, so a caller can compare without sending a digest. - New:
create_mediarefuses bytes that are not a whole image of the declared type withinvalid_image: PNG chunk checksums and end markers are checked, not only the header. Sizes WordPress expected but could not make are a warning rather than an empty list. Its errors are now JSON text with a code and details. - New:
media_upload_start,media_upload_chunkandmedia_upload_finishupload an image over about 3 KB in pieces, with a digest on the whole file and optionally on each piece. The result is whatcreate_mediareturns. Retrying a chunk or a finish is safe. Unfinished uploads are removed after an hour with no activity. - New:
get_custom_cssandupdate_custom_cssread and set the theme’s Additional CSS, stored exactly as sent. CSS that loads anything from another site is refused, and a person can still add that in the Customizer. Administrators only, with a WordPress user behind the credential. Rollback restores the previous CSS for the theme it was written for. - Changed: The post that holds the theme’s Additional CSS can no longer be used as a target of other write tools such as
update_postandupdate_post_meta(custom_css_target). Useupdate_custom_css. - New:
update_user_profilechanges a user’s display name, nickname, first and last name, description and website. Email, password and roles cannot be changed. The display name must be one WordPress offers for that user, or the call must set the nickname to the same value. - Fix: Changing the site title or tagline through
update_site_settingssaved the value but returned an error and no receipt on a site with Elementor, when the site key had no owner or Elementor’s kit was missing or in the trash. It now returns the receipt as usual, and rollback works the same way. - Fix: With a site key that had no owner, Elementor’s Site Settings kept the old title or tagline after such a change, and publishing Site Settings wrote it back. It now follows the change.
1.18.0
- New:
update_site_settingssets the site icon:site_icontakes a Media Library attachment ID, and 0 clears it without deleting the attachment. The image must be PNG, JPEG, GIF or WebP and square; it is not cropped, and under 512×512 it is set with a warning. The response gives before and after with ID and URL,icon_url_32andicon_url_192, andwarnings. It hasdry_runand rollback, including back to no icon, and purges the whole site. - New:
update_site_settingscreates the sizes WordPress uses for a site icon, and only those, solink rel="icon"points at a file of the right size. Sending the ID that is already set creates any that are missing, which repairs an icon set through Elementor’s Site Favicon. - New:
update_site_settingssets the theme’s logo:custom_logohas the same checks, except that it does not have to be square. WordPress keeps a logo per theme, so its rollback is refused after a theme switch. - New:
get_site_settingsreturnssite_iconandcustom_logo, each{id, url}or null. A nullurlmeans the image no longer exists. - New: When Elementor’s Site Settings has a saved favicon or logo, a change through
update_site_settingsor its rollback updates it too. Publishing Site Settings otherwise writes the old image back over the new one. - Fix: Rolling back
create_mediafor the image used as the site icon or logo is now a conflict needingforce, which deletes the image and clears the setting. It left the site pointing at an image that no longer existed.
1.17.0
- New:
create_elementor_templatecreates a template the way Add New Template does: header, footer, single post and page, archive, search results, 404, popup, section, container or page. Theme Builder types need Elementor Pro. It builds a draft, fills it, sets its conditions and publishes last, so a failure on the way leaves nothing showing. A header or footer with conditions cannot be published empty, because it would replace the theme’s with nothing. Administrators only. - New:
set_elementor_template_conditionssets where a template shows, through Elementor Pro’s own conditions manager, which rebuilds the cache that decides what renders. It returns before, after andverified, and warns when another published template may show on the same pages. The same conditions again rebuild the cache if it has gone stale. Administrators only. - New:
create_menucreates a navigation menu, optionally in a theme location, andassign_menu_locationputs a menu in a location or clears it. Rollback ofcreate_menudeletes the menu and gives the location back; a menu that has items needsforce, because deleting a menu deletes its items for good, and one given a location afterwards needs that change rolled back first. - New:
get_menuslists every location the theme registers, with the menu in it (locations). - New:
resolve_urlsays which Theme Builder header and footer a URL gets (theme_locations), by Elementor Pro’s own rules applied to that URL. A page on Elementor Canvas gets neither, and says so (hidden_by_page_template). Another host gives null, and what it cannot judge goes undernot_evaluated. - New:
list_elementor_templatesreturnsconditions_revision, forif_revision. - New: Server instructions describe the header and footer workflow, menus, and
elementor_header_footer. - Fix:
resolve_urlandget_postwithinclude_shadowingreported a header, footer, popup or section on Entire Site as what renders every page, so a request to edit a page could end up editing the site header. Only templates that replace a page’s content count now. - Fix:
resolve_urlfollows Elementor Pro’s rules for the blog, archives, the front page and author archives. A blog archive template was reported as shadowing ordinary pages. - Fix:
resolve_urldecides what kind of page a URL is from WordPress’s own rewrite rules and query, instead of guessing from its path. Later pages of the blog and of archives, pretty search URLs, custom taxonomies and plain-permalink query strings were read as not found or as the front page, and terms and authors that do not exist as archives.route_typegainstaxonomy_archive,date_archive,search_resultsandblog_homefor pages it called404. - Fix: A singular template is no longer reported as shadowing the blog index or a search.
- Fix:
update_postrefuses to publish an empty header or footer that has display conditions, which would remove the theme’s from those pages. - Fix: On a post with Elementor off,
update_postno longer says a content edit will not reach visitors. - Fix: The result of Save Settings now appears next to the button. It appeared at the top of a long page, out of sight, and a failed save looked the same as a successful one.
1.16.0
- Security: Hardened which stored data a post meta write can reach, and how a rollback treats stored data. Update recommended.
- Change: update_post_meta refuses a key the database would read as a different key already stored on the post or, for a key new to the post, as another spelling stored on the site (meta_key_mismatch). A new key with spaces or characters outside ASCII needs force. A key over 255 characters is refused (meta_key_too_long); nothing was stored before, though the response said it was.
- Fix: Changing a page’s title, status, slug or excerpt with update_post no longer removes iframes and scripts from content the call did not mention. Rolling back a post field now restores it as it was stored; the same filter used to remove them from the restored value, and rollback still reported success.
- Fix: An Elementor element tree that would crash the page for visitors, such as a JSON object or an element with no elType, is refused with elementor_invalid_tree. It used to be stored and reported as a success.
- Fix: A revision’s ID can no longer be used as a write target (revision_target). WordPress sent the change to the parent post, past the checks made on it.
- Fix: Rolling back an Elementor write on a page with no elements yet now restores the content visitors saw, also when Elementor’s flag was on.
- Fix: Rolling back “Elementor turned off” after the page was edited is now a conflict, because turning it back on would hide that edit.
- Fix: update_seo_data, and its rollback, no longer damage SEO plugin variables such as %%category%% and %%date%%, or encoded characters in a canonical URL when it is read.
- Fix: A widget that fails while rendering can no longer corrupt the response of the tool call, which lost the change_id of a write that had happened.
- Fix: A write could return a change_id that was never saved, so rollback could not find it. This happened on healthy sites when a meta key was longer than 100 characters, and when a database column was missing. Long keys now save. If an undo record really cannot be saved, the response says so, with no change_id, and Diagnostics shows it.
- Fix: Elementor writes no longer overwrite the content of a page sent “Back to WordPress Editor”. They are refused unless update_elementor_data is called with activate_builder: true, and rollback then restores the content, the setting and the old layout. get_page_builder reports these pages with elementor_tree_present.
- Fix: A new page given a layout with update_elementor_data now renders as an Elementor page.
- Fix: Rolling back an Elementor change now also updates post_content, which search and feeds read.
- Fix: A stripped style or script no longer leaves its CSS or code on the page as visible text (WordPress 6.5 or later).
- Change: The inline style block setting checks CSS more strictly. An independent review, with Chrome, found ways past the old check to load files from other sites. A style block is now dropped whole if its text names another site, including in a // comment or an attribute selector, and a url() pointing at this site needs a path.
- New: create_post, update_post, update_seo_data, assign_term, create_term, update_term and delete_term now honour dry_run. They used to ignore it and really write. A tool with no dry run now refuses the flag with dry_run_unsupported instead of running.
- Fix: delete_term no longer reports the default category as deleted when used with a key that has no owner.
- New: Diagnostics shows database table problems and undo records that could not be saved. Table problems are retried automatically.
1.15.1
- Security: Hardened how Elementor’s cached page data is handled. Update recommended. Elementor rebuilds each page once on its next view, and page caches are purged once shortly after the update.
- Change: get_post_meta no longer returns _elementor_element_cache, and a write containing Elementor’s [elementor-element] cache shortcode is refused with reserved_shortcode. Read tools can still search for it.
1.15.0
- New: Changes made through MCP now purge page caches: LiteSpeed Cache, WP Rocket, W3 Total Cache, WP Super Cache, WP Fastest Cache and SiteGround Optimizer. A change to one page purges that page, plus the home URL for the front page or posts page. A change to the Elementor kit, Theme Builder templates, global widgets, saved templates, menus, terms, redirects or site settings purges the whole site. Turn it off under Settings > IATO MCP.
- New: Write tools return a cache object (detected, purged, scope, notes) saying what was purged and naming caches IATO cannot reach, such as Cloudflare or a host’s CDN. get_site_info lists what it found. Dry runs and reads have no cache object.
- New: update_site_settings sets the site title, the tagline and what the home page shows. It accepts dry_run, and each option changed has its own rollback. get_site_settings also returns show_on_front, page_on_front and page_for_posts.
- New: sanitized_report says what sanitizing removed from each value: which tags and attributes, and which rule applied. create_post, update_post, the create_media description and plain text values in update_post_meta and set_page_settings report removals too; they reported nothing before. sanitized and sanitized_paths are unchanged.
- New: Setting “Allow inline SVG and style blocks in HTML widgets for administrators”, off by default, needs WordPress 6.5 or later. When on, the Elementor HTML widget keeps style blocks and a small SVG subset for a WordPress user who holds both unfiltered_html and manage_options. Scripts, event attributes, javascript: URLs, foreignObject, @import and off-site url() are still removed, and the report says what was kept and what was not.
- New: Developer hooks iato_mcp_cache_purge (filter, skip a purge) and iato_mcp_after_cache_purge (action, add your own).
- Change: Editing a classic Elementor page now clears and rebuilds only that page’s CSS, as Elementor’s own editor does. Every edit used to delete every page’s CSS file, which could leave other cached pages unstyled. Writes that other pages render still clear all of it.
- Fix: When Elementor’s own save removes content for a user without unfiltered_html, including from widgets the write did not touch, the result now says so.
1.14.0
- New: Heading level, text, link and external-image alt fixes now work on Elementor Atomic Editor (V4) elements through the existing tools (set_heading_level, set_widget_setting, update_elementor_widget, update_elementor_widgets_bulk, update_elementor_patch, update_alt_text). IATO converts and validates values with Elementor’s own code (Elementor 4.3.0 or later), and edits go live like classic edits.
- New: Each atomic change records a receipt per element and can be rolled back on its own, with the conflict check and dry_run.
- New: Results say how a change was applied (applied_via), that it is live (state), which values were sanitised, and which caches were cleared.
- New: update_alt_text accepts post_id and element_id to fix the alt text of an atomic image that uses an external URL. For a Media Library image it changes the attachment’s alt text, as before. update_alt_text also accepts dry_run.
- Change: Plain values sent to atomic elements, refused since 1.13.1, are now converted and validated with Elementor’s code and stored in Elementor’s format. On sites without Elementor 4.3.0 or later, the 1.13.1 rule stays: only typed values are accepted, and the refusal (now atomic_write_unavailable) says what is missing.
- Change: While Elementor cannot load an atomic element’s type (the Atomic Editor is off, or the plugin that provides it is inactive), IATO no longer changes that element, because nothing can validate it; typed values were stored unchecked until now. Other elements on the page can still be edited.
- Change: In update_elementor_widgets_bulk, atomic rows for the same page are applied in one save, so several rows that send the same if_revision now succeed together. Before this change, every row after the first failed with revision_conflict. Classic rows are unchanged.
- New: get_iato_seo_fixes lists issues it cannot act on under skipped, with a reason: a URL that is not a WordPress post on this site, or a URL on another domain.
- Fix: An error thrown inside a tool or another plugin now returns a clear error instead of a PHP fatal error.
- Fix: When Elementor’s own save fails and IATO stores the page itself, IATO clears that page’s CSS and element cache and says so in the result.
- Fix: Sending a setting again to an atomic element that an earlier IATO version stored as a plain value repairs that element.
- Tweak: Atomic edits clear only the edited page’s caches, not every page’s.
- Tweak: Wording of the plugin description.
1.13.1
- Fix: plain values sent to Elementor Atomic Editor (V4) elements are now refused instead of being stored in a form that could remove links or images and block later saves.
- Fix: IATO no longer stores new settings for an Atomic Editor (V4) element that Elementor’s own validation rejects, whichever credential is used, and dry runs report the same refusal. Settings an atomic element does not have (such as header_size on an atomic heading) are refused instead of being reported as applied.
- Fix: editing a page that contains elements from a deactivated addon, or atomic elements while the Atomic Editor is off, no longer removes those elements.
- New: the Elementor read tools (get_elementor_data, get_elementor_widget, list_elementor_widgets, find_elementor_widgets) flag an atomic element whose stored settings Elementor cannot read with stored_value_invalid and invalid_keys, and show those values as stored. See the FAQ for how to repair such an element.
1.13.0
- Security: OAuth connections (the Claude connector, Claude Code, Cursor and other MCP clients) now receive their own tokens. The authorization code is random and single use (5 minutes); the access token expires after 1 hour and is renewed with a rotating refresh token (30 days); a connection expires after 180 days unless it is approved again (filter: iato_mcp_oauth_grant_lifetime). Tokens are stored hashed. Before this release the OAuth code and the access token were the site’s API key itself.
- Security: each OAuth connection acts as the administrator who approved it and is re-checked on every request. Deleting or demoting that user revokes the connection. Approving a connection no longer assigns the site’s API key to the approving user.
- New: Settings > IATO MCP lists connected OAuth clients (application, approving user, connected, last used) with a Revoke button per connection and a Disconnect all button, and the plugin serves a token revocation endpoint at /oauth/revoke. Revoking connections does not affect the API key.
- Security: dynamic client registration accepts https redirect URIs only (plain http only to localhost, 127.0.0.1 or [::1], any port), stores the application name, is rate-limited (10 per IP per 10 minutes, 100 per site per hour; filter: iato_mcp_oauth_registration_limits), and removes registrations that never complete an authorization after 24 hours or have had no live connection for 30 days.
- Security: PKCE (S256) is required at /oauth/authorize; the resource parameter (RFC 8707) is checked; the authorization response carries iss (RFC 9207); the consent screen shows the application name, the host it will send you back to, and the user the connection will act as, and its approval is bound to that exact request. Logged-in users who are not administrators get a clear “administrators only” page. Follows the MCP authorization specification revision 2026-07-28. Client ID Metadata Documents (“Use Claude’s published identity”) are not supported yet; Claude falls back to dynamic registration …
