Description
Trakkr Crawler Tracking records requests whose User-Agent matches a known AI crawler pattern. It runs on the server after WordPress finishes the request, so it can record the final status code without adding a script, pixel, cookie, or public credit to your site.
The plugin is designed for site owners who use the Trakkr service. After you connect WordPress to Trakkr with a WordPress Application Password, Trakkr can retrieve recent crawler records through authenticated REST endpoints and acknowledge the rows it has received.
It also includes AI Pages, which is off by default. If your authenticated Trakkr connection turns it on, known AI crawlers can receive a version of a page from Trakkr. People visiting your site always get your normal page.
What it does
- Matches known AI crawler and AI assistant User-Agents, including GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Agent, and others.
- Stores the crawler name, User-Agent, requested URL, request method, referring URL, response status, IP address, country code, and timestamp.
- Removes query strings and fragments from requested and referring URLs before storage.
- Supports common proxy headers from Cloudflare, Sucuri, and other reverse proxies. These IP values are informational and are never used for access control.
- Exposes records only through capability-protected WordPress REST endpoints.
- Adds a dedicated Trakkr Crawler Sync role that can retrieve and acknowledge crawler rows and turn AI Pages on or off, but cannot publish content, edit posts, upload media, change other site options, or write files.
- Lets an authenticated site administrator publish approved robots.txt, llms.txt, llms-full.txt, sitemap redirect, structured data, and supported SEO metadata changes from Trakkr.
- Deletes synced local rows after 30 days and unsynced local rows after 90 days.
- Optional, and off by default: AI Pages asks Trakkr for a version of a page when a known AI crawler requests it, and tells Trakkr when a published post is saved. The External Service section lists exactly what is sent.
With AI Pages off, the plugin makes no outbound requests. It adds no remote fonts, tracking pixels, or third-party scripts to the pages people see. It does not include a custom updater. WordPress.org supplies plugin updates for the directory release.
External Service
This plugin works with Trakkr, an AI visibility service operated by Trakkr. A Trakkr account is required to view the collected crawler records in the Trakkr dashboard, use Trakkr’s connected-site publishing features, or use AI Pages.
Requests from Trakkr to your site
When an authorized user explicitly connects the site in Trakkr, the Trakkr service sends authenticated HTTPS requests to this site’s WordPress REST API. WordPress Application Password authentication and WordPress capabilities control access. Crawler-only connections can use a dedicated user with the Trakkr Crawler Sync role. Existing administrator connections remain supported.
For crawler syncing, the response can contain the crawler User-Agent, crawler name and type, requested URL, request method, referring URL, IP address, country code, response status, and timestamp. Query strings and fragments are removed from both URLs. Trakkr uses this data to show crawler activity for the connected site. Trakkr also sends the IDs it received back to WordPress so those local rows can be marked as synced.
If the administrator uses connected-site publishing, Trakkr can read or update only the allow-listed WordPress options, files, post metadata, and publish-operation markers exposed by this plugin. These actions occur only after the administrator connects the site and requests or approves the related action in Trakkr.
Requests from your site to Trakkr (AI Pages)
AI Pages is off by default, and while it is off the plugin makes no outbound requests. It only starts after an authenticated Trakkr connection turns it on and stores a key through the trakkr/v1/ai-pages REST route. That connection must use an Administrator or a user with the Trakkr Crawler Sync role.
Once it is on, the plugin sends HTTPS requests to one host, prism.trakkr.ai, in two cases.
- When a known AI crawler requests a page. The plugin asks Trakkr whether it has a version of that page for the crawler. It sends the page URL including any query string, the URL path, the name of the matched crawler, the first 200 characters of the crawler’s User-Agent, your site’s AI Pages key, and the plugin’s AI Pages version. It waits at most 1 second. Anything other than a usable answer means the crawler gets your normal page.
- When a published post is saved. The plugin tells Trakkr the page changed. It sends the post’s public URL, the reason
cms_save, your site’s AI Pages key, and the plugin’s AI Pages version. It does not wait for an answer. Drafts, autosaves, revisions, password-protected posts, and post types with no public page send nothing.
WordPress also adds its usual HTTP User-Agent, which names your WordPress version and your site’s address. The plugin does not send the crawler’s IP address, cookies, or the content of your posts.
If Trakkr answers with a version of the page, the plugin sends that HTML to the crawler that asked, marks the response so caches do not keep it, and stores nothing. The crawler visit is still recorded locally as usual.
No request is made for browsers and other User-Agents that do not match, logged-in users, wp-admin, AJAX, the REST API, cron, feeds, previews, missing pages, non-GET requests, or static files.
A User-Agent matches if it contains any of these, in any letter case: chatgpt-user, claude-user, perplexity-user, mistralai-user, meta-externalfetcher, oai-searchbot, claude-searchbot, perplexitybot, gptbot, claudebot, meta-externalagent, amazonbot, cohere-ai, mistral, deepseek, grok, bytespider, ccbot, bingbot, applebot, google-agent. The plugin asks Trakkr for all of these. Your settings in Trakkr decide which ones receive a version of the page.
To stop these requests, turn AI Pages off in Trakkr, or use the “Turn off AI Pages” link on the WordPress Plugins screen, which needs no Trakkr account and is shown to Administrators while AI Pages is on. Deactivating or deleting the plugin also stops them. AI Pages stays off unless a Trakkr connection turns it on again, and revoking that connection’s Application Password rules that out.
- Trakkr service: https://trakkr.ai
- AI Pages service host: https://prism.trakkr.ai
- Terms of Service: https://trakkr.ai/terms
- Privacy Policy: https://trakkr.ai/privacy
Privacy
The plugin writes a local database row only when a request’s User-Agent matches one of its known AI crawler patterns. User-Agent strings can be imitated, so site owners should still treat the stored IP address and URL paths as potentially personal data.
To reduce collection, query strings and fragments are removed from requested and referring URLs before storage. The plugin does not set cookies and does not add front-end JavaScript.
If AI Pages is turned on, the plugin also sends data to Trakkr for matching crawler requests and for saves of published posts, as listed under External Service. That includes the page URL with its query string, because the query string can select a different page. Those requests are not stored in the local crawler table. People visiting your site are not involved: their requests are never sent to Trakkr by this feature.
Synced rows are deleted from the WordPress database after 30 days. Rows that have not been synced are deleted after 90 days. Deleting the plugin removes its database table, scheduled cleanup event, stored options including the AI Pages key, and its role and capabilities.
Trakkr keeps connected-site data according to its Privacy Policy. Site owners should review that policy and add the suggested Trakkr text from Settings > Privacy to their own privacy policy where applicable.
Installation
- In WordPress, go to Plugins > Add New.
- Search for “Trakkr Crawler Tracking”, then install and activate it.
- In Trakkr, open your site’s crawler tracking connection and choose WordPress.
- For crawler-only access, create a dedicated WordPress user with the Trakkr Crawler Sync role. Use an Administrator only if you also want Trakkr’s publishing and site-management features.
- Create an Application Password for that user and enter it in Trakkr over HTTPS.
- Confirm the connection in Trakkr. Tracking begins on the next request whose User-Agent matches a known AI crawler pattern.
- AI Pages is optional and off by default. To use it, turn it on in Trakkr. There is nothing to upload or paste into WordPress.
No WordPress settings screen is required. The crawler feed requires the dedicated trakkr_read_crawler_visits and trakkr_ack_crawler_visits capabilities. The plugin grants both to its Trakkr Crawler Sync role and to Administrators for backwards compatibility. The AI Pages route requires manage_options or the dedicated trakkr_manage_ai_pages capability, which the plugin grants to the same two roles and which allows nothing else. Site-wide file and option endpoints still require manage_options. Post metadata endpoints still verify that the authenticated user can edit the specific post.
FAQ
Can Trakkr edit Slim SEO fields?
Version 1.7.0 adds a versioned SEO bridge for Slim SEO. Update this plugin, then use Check connection in Trakkr’s connected-site settings. Trakkr writes titles and descriptions to Slim SEO’s native settings and merges approved schema changes into its existing graph. Other settings and schema relationships are preserved. Disabled Slim SEO features are not advertised as writable. Existing Yoast, Rank Math, SEOPress, and AIOSEO connections keep their existing paths.
SEO access requires permission to edit the specific post; crawler-only accounts cannot use it. Undo restores the previous field only if it has not been changed since. Use Trakkr’s undo before removing the plugin if you want to revert an applied change. Removing the bridge stops its schema additions, while saved native titles and descriptions remain in Slim SEO.
The trakkr_seo_providers PHP filter is an extension point for explicitly supported providers, not automatic support for every SEO plugin. If multiple providers are registered, the bridge refuses to choose one.
Does this slow down my site?
The matching work is a short in-memory pattern check on each WordPress request. Only a matching request triggers a database insert. Detection runs on WordPress’s shutdown hook after the response has been generated.
AI Pages is off by default. When it is on, it adds a request to Trakkr only for matching crawlers, with a hard ceiling of 1 second. If Trakkr is slow or unreachable, that crawler gets your normal page. People visiting your site never wait on it. Saving a published post sends a short notice at the very end of the request, after your host has returned the response where it supports that (PHP-FPM and LiteSpeed). On other hosts it adds at most 1 second per notice, with no more than 3 notices in one request.
Does AI Pages change what people see?
No. Only requests whose User-Agent matches the crawler list above can receive a version of a page from Trakkr. Those responses tell caches not to keep them, so a page cache or CDN should not hand one to a person. Googlebot is not on the list. Bingbot, Applebot, and Google-Agent are, and your settings in Trakkr decide whether they receive a version of the page.
How do I turn off AI Pages?
While it is on, Administrators see a “Turn off AI Pages” link beside this plugin on the WordPress Plugins screen. It takes effect straight away and needs no Trakkr account. You can also turn it off in Trakkr, or deactivate the plugin. It stays off unless a Trakkr connection turns it on again. To rule that out, revoke the connection’s Application Password under Users > Profile.
Does it track human visitors?
The plugin only records requests whose User-Agent matches a known AI crawler pattern. Regular browser User-Agents are ignored. Because User-Agent strings can be imitated, a matching row is evidence of the claimed crawler identity, not proof that the request came from that company.
What data is stored in WordPress?
The wp_trakkr_crawler_visits table stores the crawler User-Agent, crawler name and type, requested URL, request method, referring URL, response status, IP address, country code, timestamp, and sync state. Query strings and fragments are removed from both URLs. The wp_ part follows your site’s configured database prefix.
Does Trakkr need an administrator Application Password?
No, not for crawler tracking or AI Pages. Use a dedicated account with the Trakkr Crawler Sync role. That role can retrieve crawler rows, mark received rows as synced, and turn AI Pages on or off, but cannot publish content, edit posts, upload media, change other site options, or write files. An Administrator or suitably privileged Editor is needed only for the separate connected-site publishing features. Keep each Application Password only in Trakkr and revoke it in WordPress when you disconnect the site.
Does it work behind Cloudflare or Sucuri?
Yes, for matching requests that reach WordPress. The plugin can read standard proxy headers to record the reported client IP and country. Requests served entirely from a CDN or blocked before reaching WordPress cannot be recorded by an origin plugin.
Does crawler tracking work on WordPress VIP?
The crawler endpoints and database records are compatible with WordPress VIP’s read-only application containers. The plugin stores llms.txt content in the WordPress database and serves it through WordPress, so it does not write to the application filesystem. WordPress VIP customers must add the unpacked plugin through their application Git repository and should validate performance in a non-production environment before enabling it in production.
Does the plugin download or execute remote code?
No. It contains no remote-code loader and no custom update service. The WordPress.org directory provides updates for this release. The only thing it receives from Trakkr is the HTML of a page when AI Pages is on. It passes that HTML to the crawler that asked and never stores or runs it.
Screenshots
Changelog
1.9.0
- Add AI Pages, off by default. An authenticated Trakkr connection can turn it on, with no file to upload. Known AI crawlers can then receive a version of a page from Trakkr. People always get the normal page.
- This release can make outbound requests: only to prism.trakkr.ai, only once AI Pages is on, with a 1 second ceiling when serving a crawler. Any failure serves the normal page. See External Service.
- Add a “Turn off AI Pages” link on the Plugins screen for Administrators, which needs no Trakkr account.
- Add the
trakkr_manage_ai_pagescapability to the Trakkr Crawler Sync role and Administrators. It allows the AI Pages route and nothing else. - Add AI Pages status to the crawler connection’s status response, and disclosure text to Settings > Privacy.
1.8.0
- Report field-level SEO support, disabled features, and conflicting providers.
- Add opt-in titles and descriptions for compatible public pages without an SEO provider. Existing SEO output is left alone; turning this off restores the site defaults. Requires the WordPress HTML API with text editing support.
- Protect schema entity identities when an edit supplies a different ID for the same subject.
1.7.0
- Add a versioned SEO bridge with native Slim SEO title and description updates, schema graph integration, and field-level conflict checks for undo. Existing SEO integrations and crawler-only permissions are unchanged.
1.6.1
- Store llms.txt and llms-full.txt content only in prefixed WordPress options and serve it through WordPress.
- Removed all web-root file reads, writes, and deletions.
- Replaced dynamic option API calls with fixed prefixed option names.
- Prefixed the PHP 7.4 list-array compatibility helper.
1.6.0
- Added the least-privilege Trakkr Crawler Sync role for crawler-only connections.
- Split crawler read and acknowledgement permissions from the broad
manage_optionscapability. - Kept existing Administrator connections working by granting Administrators the new crawler capabilities during upgrade.
- Added multisite-aware activation and uninstall cleanup for the crawler table, role, capabilities, and cleanup schedule.
- Skip optional physical llms.txt writes on read-only application filesystems such as WordPress VIP.
1.5.0
- Prepared the plugin for WordPress.org directory distribution and WordPress-managed updates.
- Updated compatibility metadata through WordPress 7.1.
- Added complete external-service, privacy, retention, authentication, and uninstall disclosures.
- Removed query strings and fragments before storing requested and referring URLs.
- Replaced Host-header URL construction with the configured WordPress home URL.
- Limited acknowledge requests to 5,000 unique visit IDs.
- Added suggested text to the WordPress privacy-policy guide.
- Migrated physical-file ownership markers from MD5 to SHA-256 while retaining safe upgrade compatibility.
- Added translated REST error messages and explicit text-domain and GPL headers.
1.4.2
- Content publishing now rejects ambiguous duplicate operation markers instead of choosing one.
1.4.1
- llms.txt and llms-full.txt can be written as physical files in the web root for hosts that serve text files without reaching WordPress.
- Files are re-created on plugin activation and update. Files not created by this plugin are never overwritten or deleted.
1.4.0
- Added validated JSON-LD output from Trakkr schema metadata.
- Added llms-full.txt support and an optional same-site sitemap redirect.
1.2.0
- Moved detection to the
shutdownhook so recorded status codes are accurate. - Added request method, referrer, country, proxy IP support, schema migrations, retention, uninstall cleanup, and administrator-only crawler sync.
1.1.0
- Added authenticated robots.txt and llms.txt management plus publish-operation lookup.
1.0.0
- Initial release.


