Odyssey LLMS

By odysseynewmedia

★ 5.0/5 (3 ratings)300+ active installationsVersion 6.4.12

The complete AI control panel for WordPress. Generate llms.txt, Markdown, and JSONL. Includes AI Analytics and Content Cleaning.

Description

Odyssey LLMS is the definitive control panel for your website’s AI presence. It generates the critical files used by Large Language Models (like ChatGPT, Claude, and Gemini) to understand and cite your content.

For Beginners: Just activate it. A comprehensive and optimised llms.txt file is instantly generated. No configuration is needed.

For Power Users: Manage every aspect of your AI strategy. Track bot traffic with built-in analytics, generate JSONL datasets for fine-tuning, and clean up your content with CSS selectors.

Concepts Explained: Why do you need this?

1. What is llms.txt?

Think of llms.txt as a “Sitemap for AI”. While humans use HTML pages and Search Engines use XML sitemaps, AI agents look for an llms.txt file in your root directory. This file gives them a clean, prioritised list of links to crawl, ensuring they train on your best content and ignore the junk.

2. What is llms-full.txt (Markdown)?

This is an optional advanced feature (RAG-Ready). Instead of just providing links, llms-full.txt provides your actual website content converted into clean, lightweight Markdown format.
* Why it’s useful: It allows AI agents to ingest your website’s knowledge immediately without needing to visit and scrape every single HTML page. This reduces server load and ensures the AI gets accurate data for “Retrieval Augmented Generation” (RAG).
* ⚠️ WARNING regarding Virtual Mode Limits: When using Virtual Mode to generate this file, the item limit for the llms-full.txt file is securely capped at 50 by default. Manually increasing this limit beyond 50 in the ‘Tools’ settings will drastically increase server load and risks causing immediate 500/503 server crashes. Use this feature at your own risk. If you require more than 50 items in your llms-full.txt file, we recommend using Physical Mode instead.

3. What is llms.jsonl (Fine-Tuning)?

This file formats your content into prompt-completion pairs (JSON Lines). This is the standard format used to fine-tune models like GPT-4 or Llama 3 on your specific data.

New Features in 6.0:

  • JSONL Generator: Create a dataset ready for fine-tuning custom AI models.
  • Content Cleaning: Use CSS selectors (e.g., .sidebar, .comments) to strip unwanted elements from your Markdown and JSONL files.
  • Visual Analytics: Visualise exactly which AI bots (ChatGPT, Claude, Google AI, etc.) are accessing your files with a built-in dashboard widget and charts.
  • WooCommerce Integration: Automatically generates a structured “Products” section including Price, Stock, and SKU. Large catalogues are generated in small background batches, so big stores do not time out.

Key Features:

  • Clean Tabbed Interface: Organised into General Rules, Content Sourcing, Analytics, Robots.txt, Security, and Tools.
  • Granular Bot-Specific Rules: Set detailed Allow or Disallow rules for individual AI crawlers (GPTBot, Google-Extended, etc.).
  • Settings Import & Export: Perfect for agencies. Easily back up, restore, and migrate your settings between sites with JSON import/export.
  • Advanced Scheduling: Regenerate your file on save, manually, or on a recurring schedule (Hourly, Daily, Weekly, Monthly).
  • Safety Validator: Prevents accidental blocking of all traffic in robots.txt.
  • Agentic Resource Discovery (ARD): Publishes /.well-known/ard.json so AI agents and registries can discover your llms.txt, llms-full.txt and llms.jsonl files (Google PageSpeed Insights ARD check).
  • Multilingual Ready: Works with WPML and Polylang. Pages from every active language are included, each with its correct translated address, whichever language you are viewing the admin in.

Third-party libraries

Human-readable sources ship alongside minified assets under assets/vendor/:

  • Select2 4.0.13 (MIT): https://github.com/select2/select2
  • Chart.js 4.5.1 (MIT): https://github.com/chartjs/Chart.js

Installation

  1. Upload the odyssey-llms folder to the /wp-content/plugins/ directory.
  2. Activate the plugin through the ‘Plugins’ menu in your WordPress dashboard.
  3. That’s it! A llms.txt file has been generated. A new “Odyssey LLMS” menu item will appear in your admin sidebar.
  4. (Optional) Navigate to the Odyssey LLMS page to view Analytics or configure advanced settings.

FAQ

Where do I find the settings page?

The settings page is located in its own top-level menu in your WordPress admin sidebar, labelled Odyssey LLMS.

How do I clean up the Markdown content?

In the “Content Intelligence” tab, look for the “Content Cleaning (CSS)” field. Enter CSS selectors for elements you want to remove, separated by commas (e.g., .footer, .nav, #sidebar).

How do I enable Analytics?

Go to the main settings tab and select “Virtual File” as your serving method. This allows WordPress to intercept bot requests and log them to your dashboard.

Does this conflict with Yoast SEO or Rank Math?

No. The Robots.txt editor includes intelligent conflict detection. It will automatically fetch and import any virtual rules created by other SEO plugins so you don’t lose them. Odyssey LLMS has been tested with Rank Math and respects noindex settings from both Yoast SEO and Rank Math.

Can I increase the limit of pages in llms-full.txt?

⚠️ WARNING: While you can manually increase the post/page limit for the llms-full.txt file in the settings, we strongly warn against setting it too high (especially in Virtual Mode). Doing so will drastically increase server load and risks causing immediate 500/503 server crashes due to the heavy processing required. If you require a large number of items, we recommend using Physical Mode.

What is the JSONL Prompt Template and how do I use it?

The JSONL Prompt Template controls the “question” side of each entry in your llms.jsonl fine-tuning dataset. When you want to train a custom AI model (such as GPT-4, Llama 3, or any fine-tuneable LLM) on your website’s content, that process requires structured “Question and Answer” pairs – known as prompt and completion.

How it works:

The plugin generates llms.jsonl where every line is a JSON object in this format:

{"prompt": "What is My Post Title?", "completion": "The full text of that post..."}

The template field lets you define the prompt structure for every post on your site. Use {{title}} as a dynamic placeholder. The plugin will automatically replace it with each post’s actual title at generation time.

Examples by use case:

  • Default / General: What is {{title}}?
  • Service business: Tell me about the {{title}} service provided by our company.
  • E-commerce / Products: What are the features and specifications of {{title}}?
  • Knowledge base / Support: How do I {{title}}?

By customising this template, you are priming your fine-tuning dataset so the resulting model learns to respond to the exact style of queries your users will ask.

What is the difference between Physical File and Virtual File mode?

Physical File (Recommended for most sites): The plugin writes a real static llms.txt file to your server’s root directory. This is served directly by your web server (Apache/Nginx) with maximum speed, and does not require WordPress to load for every bot request. The downside is that bot visits cannot be tracked for Analytics.

Virtual File (Enable Analytics): The file is served dynamically by WordPress via a rewrite rule. This allows the plugin to intercept every request and log it to the Analytics dashboard. It also enables Rate Limiting to throttle abusive bots. The trade-off is a small performance overhead since WordPress must load for each request.

Both modes support llms.txt, llms-full.txt (Markdown), and llms.jsonl generation.

What does “Include Taxonomy Archives” do?

When you select specific taxonomies (e.g. Categories, Tags, WooCommerce Product Categories) in the Content Source tab, the plugin will fetch all non-empty terms for those taxonomies and include their archive page URLs in the generated llms.txt. This is useful for giving AI crawlers a complete picture of your site’s topic structure, not just individual posts.

What does “Include Author Archives” do?

When this option is enabled, the plugin fetches all WordPress users who have at least one published post and appends their author archive URLs (e.g. yoursite.com/author/name/) to the generated llms.txt. This is useful for sites where author credibility and profiles are part of the content strategy.

How does the WooCommerce integration work?

When WooCommerce Products is enabled in the Content Source settings and WooCommerce is active, the plugin automatically enriches product entries with structured metadata pulled directly from WooCommerce:

  • Price – the product’s formatted sale/regular price.
  • SKU – the product’s stock-keeping unit identifier (if set).
  • Stock Status – whether the product is In Stock or Out of Stock.

This data is prepended to the product content in both llms-full.txt (Markdown) and llms.jsonl, giving AI models accurate, up-to-date product information without needing to scrape individual product pages.

What is the Content Order setting?

The Content Order setting in the Content Source tab controls the order in which posts appear across all generated files (llms.txt, llms-full.txt, llms.jsonl). Since AI agents typically give more weight to content they encounter earlier in a file, ordering by your most recent or most engaged-with content ensures they train on your best material first.

Available options:

  • Date: Newest First (Default) – Most recently published content appears first.
  • Date: Oldest First – Chronological order from the beginning.
  • Title A → Z – Alphabetical by post title.
  • Most Commented First – Prioritises your most engaged content.
  • Menu/Page Order – Uses the WordPress page order field, ideal for structured sites.

How does rate limiting work?

When Virtual File mode is active, the plugin tracks the number of requests made by each IP address within a rolling time window. If a bot or visitor exceeds the configured request limit, they are temporarily blocked and receive a 429 Too Many Requests response with a Retry-After header. The block duration is configurable in the Security tab and enforces a minimum of 1 hour to prevent configuration errors from causing permanent lockouts.

What happens to my data when I uninstall?

Uninstalling Odyssey LLMS removes scheduled jobs, temporary caches, rate-limit blocks, and any physical llms.txt / llms-full.txt / llms.jsonl files from the site root. Plugin settings and the analytics database table are preserved so you can reinstall later without reconfiguring. To fully remove analytics data, delete the custom table from your database after uninstall (or contact your host for assistance).

Does this plugin call external services?

No. Odyssey LLMS does not phone home or send your content to third-party AI APIs. Optional sitemap fetching only contacts URLs you configure (or auto-detect on your own site). Bundled admin libraries (Select2, Chart.js) are included locally. See “Third-party libraries” below.

What is the ARD manifest (/.well-known/ard.json)?

Agentic Resource Discovery (ARD) is an open specification for advertising agentic resources. When enabled (on by default), Odyssey LLMS serves a JSON manifest at /.well-known/ard.json that describes your llms.txt file and, when enabled, llms-full.txt and llms.jsonl. It also adds a <link rel="ard"> tag and an HTTP Link header so crawlers and PageSpeed Insights can find it. Since 6.4.10 the plugin writes the manifest as a small static file at /.well-known/ard.json (in both Physical and Virtual mode) and refreshes it after every rebuild and settings change, because PageSpeed Insights only waits 2 seconds for the file and a full WordPress page load can be slower than that on shared hosting. If the folder is not writable, the manifest is served dynamically instead and the ARD setting shows a warning. If another plugin or you already provide /.well-known/ard.json, that file is served as-is (and still advertised); the plugin never changes or deletes a file it did not create. You can edit the representative queries used for discovery under General settings. Some hosts or caching plugins may need an exception for /.well-known/*.

Does the plugin add an Agentmap line to robots.txt?

Not by default. Since 6.4.11 the “Add Agentmap line to robots.txt” setting is off, because some SEO tools, including the current version of Google PageSpeed Insights, report Agentmap: as an unknown robots.txt directive. AI agents can still find your ARD manifest through the rel="ard" link and /.well-known/ard.json. You can tick the setting if you specifically want the line.

Does it work with WPML or Polylang?

Yes. When WPML or Polylang is active, the plugin includes your pages in every active language, grouped by language if you choose, and uses each page’s own translated address in both llms.txt and llms-full.txt. After changing translations, click Regenerate Now in the Tools tab to refresh the files.

Screenshots

Changelog

6.4.12

  • WPML and Polylang: matching page addresses in every file: Fixed some translated pages showing a different address in llms-full.txt from the one in llms.txt (for example, a French page listed under the Dutch page name). Every generated file now takes each page’s address from one shared place, worked out in the page’s own language, so they always match. Generated files are rebuilt once after the update on multilingual sites.

6.4.11

  • WPML and Polylang: all languages in one file: Fixed generation only including pages in the current admin language (or the default language during WP-Cron and background rebuilds). llms.txt, llms-full.txt and llms.jsonl now include eligible content from every active language, grouped by language when that setting is on, independently of the request language. Permalink and term queries use the same all-language context.
  • llms-full.txt missing most pages: Fixed expanded markdown skipping pages whose content lives in nested Gutenberg or Greenshift (and similar) blocks, because only the empty wrapper HTML was read. Legal/document shortcodes such as Complianz are rendered when the static HTML is empty. Eligible pages still get a title, source and summary if the body cannot be converted, so a file is no longer reduced to a single classic-editor page, or to a title and introduction with no sections.
  • “Discourage search engines” does not empty llms-full.txt: Site Visibility / blog_public is not used to drop page sections. Individual Yoast or Rank Math noindex flags are still respected when that setting is on.
  • Agentmap line off by default: The plugin no longer adds Agentmap: to robots.txt by default. Current PageSpeed Insights (Lighthouse 13.6) flags it as an unknown directive. A new “Add Agentmap line to robots.txt” setting is off by default; existing installs are switched off on upgrade, and older settings imports do not turn it back on. The rel=”ard” link and /.well-known/ard.json remain in place for discovery. Existing physical robots.txt files have the managed Agentmap block removed on upgrade.

6.4.10

  • ARD loads within PageSpeed’s time limit: Fixed “Advertised ARD manifest location could not be loaded” in Google PageSpeed Insights. Lighthouse only waits 2 seconds for /.well-known/ard.json, and building it through WordPress could take longer on shared hosting when the page cache was cold. The plugin now writes a static /.well-known/ard.json file in both Physical and Virtual mode, so the web server returns it straight away.
  • The static file is refreshed after every rebuild (manual, background and scheduled), when settings or the site address, title or tagline change, and on upgrade. A throttled check recreates it if it goes missing, so sites already on 6.4.9 get it automatically.
  • The file is removed when ARD is switched off, on deactivation and on uninstall. A /.well-known/ard.json that the plugin did not create is never overwritten or deleted.
  • If the folder is not writable, the plugin falls back to serving the manifest dynamically, and the ARD setting and File Status panel show which method is in use.

6.4.9

  • Agentic Resource Discovery (ARD): Serves /.well-known/ard.json (spec v0.91) describing llms.txt and optional llms-full.txt / llms.jsonl, with a <link rel="ard"> tag, HTTP Link header, and an Agentmap: line in robots.txt. Includes an on-by-default settings toggle, editable representative queries, File Status sidebar link, import/export support, and rewrite cleanup on deactivation. The manifest is always virtual; a physical .well-known/ard.json file is left alone and still advertised when present.
  • FIX: ARD responses now support HEAD/OPTIONS, send CORS headers for browser-based agents, and avoid advertising discovery links on feeds, 404 pages and the ARD document itself.
  • FIX: Importing older settings JSON no longer turns ARD off when the export predates the feature; IDN hosts use punycode in URNs when the intl extension is available.
  • FIX: Settings import no longer triggers an undefined filter_date notice when older exports omit that key.
  • FIX: ARD REQUEST_METHOD handling passes WordPress Plugin Check input sanitisation.
  • Background regeneration loop: Fixed a bug where a leftover scheduled job could start a full rebuild of every file on each WP-Cron run (roughly once a minute on busy sites). A finished or failed job now clears its own scheduled event and is never restarted automatically.
  • False “0 URLs” warning: Fixed the empty generation notice and failed build status that could appear after first activation, “Reset to Defaults” or “Import Settings” even when llms.txt was written correctly. Virtual mode now reports the correct URL count as well.
  • Files stay available during rebuilds: Physical llms.txt, llms-full.txt and llms.jsonl are no longer deleted at the start of a rebuild, so visitors and AI crawlers keep receiving the full files while a new version is generated. Only files you have disabled are removed.
  • Consistent rebuilds: Saving settings, editing content and scheduled rebuilds now use the same chunked background job as “Regenerate Now”, which avoids timeouts on large sites and keeps the progress status accurate.
  • Deactivation: The llms.txt, llms-full.txt and llms.jsonl rewrite rules are now removed when the plugin is deactivated.
  • Robots.txt safety check: The “User-agent: ” with “Disallow: /” check now also catches “Disallow: /“, “Disallow: /*$” and rules with inline comments, both in the browser and on the server.
  • Import and Reset notices: “Settings imported successfully”, “Invalid import file” and “Settings have been reset to default” messages are now shown after the page reloads instead of being lost.
  • PHP warnings on the plugin screen: Hiding third-party admin notices on the Odyssey LLMS screen no longer triggers “Undefined array key” warnings in class-wp-hook.php.
  • PHP 8.4 and 8.5 compatibility: Analytics CSV export now passes the escape argument to fputcsv(), and CIDR matching in the IP blocklist no longer triggers a chr() deprecation notice on PHP 8.5.
  • Translations: Fixed a “translation loading was triggered too early” notice caused by the custom cron schedules, and made the “Save Robots.txt” button and “Simulated hit logged.” message translatable.
  • Code quality: Replaced current_time( ‘timestamp’, true ) with time() in the generator and logger.
  • Readme: Changelog entries for version 6.1.14 and earlier have moved to changelog.txt so the readme stays within the WordPress.org changelog length limit.

6.4.8

  • Custom AI Agents layout: Each custom agent card now uses a responsive two-row layout. Agent Name, User-Agent, Default and Remove sit on the first row and Path Rules spans the full width below, so fields are no longer truncated at common laptop widths (1280px to 1440px). The layout adapts to the space available (container queries with a media query fallback) and stacks neatly on narrow screens. The Path Rules box is now three rows tall and can be resized vertically.
  • File Status “Last Updated”: After “Regenerate Now”, the sidebar now shows the time in your site’s date and time format (for example “3 November 2026 12:52”) instead of a raw database timestamp. Regeneration results include a new timestamp_display value, and both the page render and the AJAX refresh use wp_date() so the site timezone is respected, including during British Summer Time. “Next Scheduled Run” now uses wp_date() too.
  • Screenshots: Refreshed the plugin banner and screenshots, and updated the Robots.txt caption to “Robots.txt editor with conflict detection and safety validation”.

6.4.7

  • WordPress Plugin Check Compliance: Replaced all remaining unlink() calls on temporary files with wp_delete_file() in Generator and BackgroundJob, and moved esc_url() inline inside Plugin::print_llms_discovery_links() so WordPress Plugin Check reports 0 errors.

6.4.6

  • PERFORMANCE: Replaced monolithic background generation with a chunked, resumable state machine (BackgroundJob::step()), processing llms-full.txt and llms.jsonl in time-bounded (~2.5s) batches with a shutdown fatal-error guard so large WooCommerce stores on LiteSpeed/shared hosting never time out or hang at “Writing expanded markdown…”.
  • PERFORMANCE: Eliminated double full-site generation and duplicate XML sitemap fetches during “Regenerate Now” by snapshotting existing llms.txt URLs from disk prior to build (Diff::diff_url_lists()).
  • PERFORMANCE: Added internal WordPress core sitemap resolution (wp_sitemaps_get_server()) in SitemapParser to avoid HTTP loopback requests on /wp-sitemap.xml, and restricted term_from_url() to configured taxonomies with URL path matching so WooCommerce attribute taxonomies (pa_*) are not queried on every URL.
  • BUGFIX: Replaced do_shortcode() execution in Markdown/JSONL content extraction with safe shortcode stripping and static block rendering, preventing database-heavy third-party shortcodes (Ninja Tables, TablePress, WooCommerce product grids, BeRocket filters, forms) from exhausting PHP workers and triggering HTTP 503 errors.
  • BUGFIX: Added WooCommerce product short description (post_excerpt) support in llms-full.txt and llms.jsonl and wrapped product price/metadata lookups in exception-safe handlers for headless/cron execution.
  • BUGFIX: Replaced overlapping setInterval AJAX polling with sequential setTimeout polling and transient HTTP error backoff in admin-script.js so status checks cannot pile up and saturate PHP workers.
  • PERFORMANCE: Added 'no_found_rows' => true and per-batch runtime cache/query flushing during Markdown and JSONL generation to keep memory usage flat on large catalogues.

6.4.5

  • SECURITY: Excluded password-protected posts and pages across Sitemap Mode, Priority (Must Read) URLs, and Smart Summaries so protected titles, URLs, and body excerpts are never written to llms.txt.
  • SECURITY: Blocked wildcard User-agent: * entries and validated path rule directives (Allow, Disallow, Crawl-delay) in the Custom AI Agents repeater to prevent accidental site-wide de-indexing in robots.txt.
  • BUGFIX: Switching from Physical File mode to Virtual Rewrite mode now removes existing physical llms.txt, llms-full.txt, and llms.jsonl files from disk so virtual endpoints, analytics logging, and rate limiting take effect immediately.
  • BUGFIX: Hooked virtual file serving at priority 1 on template_redirect and bypassed redirect_canonical for LLMS endpoints, fixing 301 trailing-slash redirects (/llms.txt/) on pretty permalinks and homepage rendering on plain permalinks.
  • BUGFIX: Enforced Excluded Post IDs, Excluded Categories, Date Cutoff filters, and the Author Archives toggle in XML Sitemap mode.
  • BUGFIX: Deduplicated URLs across Must Read, Pages, Posts, Blog index, Custom URLs, Taxonomies, and Author Archives so each URL appears only once in llms.txt.
  • BUGFIX: Counted the Blog index link toward the total URL count so blog-only sites no longer trigger a false “0 URLs included” warning banner.
  • BUGFIX: Fixed Gutenberg <pre><code> blocks in llms-full.txt so fenced code blocks preserve multiline formatting without nested inline backticks.
  • BUGFIX: Aligned Virtual and Physical output by removing the leading 3-byte UTF-8 BOM from physical files, applying exclude_media and the odyssey_llms_markdown_content filter in virtual llms-full.txt, and decoding WooCommerce SKUs cleanly in virtual llms.jsonl.
  • BUGFIX: Persisted last build status in a dedicated option (odyssey_llms_last_build_status) so the WordPress Site Health check remains accurate after admin notices are dismissed.
  • BUGFIX: Replaced SQL NOW() with WordPress UTC timestamps in analytics queries and cleared analytics cache transients immediately when new bot hits are logged.
  • BUGFIX: Fixed rate-limiting transient updates to preserve the initial 1-hour window expiry instead of resetting the TTL on every hit.
  • CLEANUP: Removed the deprecated “Block all unlisted bots by default” checkbox, cleaned up the # BEGIN Odyssey LLMS crawler rules block from robots.txt and plugin transients/options on deactivation/uninstall, and preserved core WordPress admin notices on the plugin settings screen.

6.4.4

  • BUGFIX: Changing the regeneration schedule now replaces the existing cron event, so Weekly and Monthly update the next run time.

6.4.3

  • ENHANCEMENT: Enabled AI crawler rules are written into a marked block in robots.txt on save. A site-wide block of every visitor is never written there.

6.4.2

  • BUGFIX: Regeneration no longer renders every page through the full front-end content pipeline, which could stall on “Writing llms.txt” and make the site return 503.
  • ENHANCEMENT: llms.txt lists pages, services, and case studies, plus one Blog link. Individual posts stay in llms-full.txt.
  • TWEAK: Crawler rules are no longer copied into llms.txt. They remain in robots.txt.
  • TWEAK: Parentheses in link titles are no longer escaped.
  • UI: WordPress Content Types is marked as the recommended content source.

6.4.1

  • ENHANCEMENT: WordPress content and XML sitemap sources now both write llms.txt links with the real page title, a content-type section, and a short note when page summaries are enabled.

6.4.0

  • COMPATIBILITY: llms.txt output now follows the llms.txt v2 format (August 2026): one H1, a blockquote summary, free-form notes, then H2 sections of markdown links with short descriptions. The Optional section is last.
  • COMPATIBILITY: Public pages advertise llms.txt with rel=”describedby” (HTML link and HTTP Link header), as described in the v2 spec.
  • ENHANCEMENT: llms.txt groups links under post-type headings (Pages, Posts, and so on) instead of one flat list. Page summaries are appended as link notes when Smart Summaries is enabled. Sitemap and WordPress content sources both use real titles, sections, and notes.
  • ENHANCEMENT: llms-full.txt uses a source line and updated date per page, and strips unrendered Divi, Fusion Builder, and WPBakery shortcodes so builder markup is not written into the file. The strip is linear so a large builder layout cannot stall generation.
  • TWEAK: Crawler Allow/Disallow rules stay in the file when enabled, but they are no longer an H2 section, so they do not break llms.txt parsers. robots.txt remains the place those rules are enforced.

6.3.3

  • COMPATIBILITY: Verified compatibility with WordPress 7.1.
  • BUGFIX: Empty-generation admin notices stay dismissed after the user clicks the dismiss control, instead of returning on later regenerations or other admin screens.

6.3.2

  • FEATURE: Automatically rebuilds or clears caches when content is published or updated.
  • FEATURE: Sitemap auto-detection for Yoast, Rank Math, and core sitemaps, plus a warning when generation finds 0 URLs.
  • FEATURE: “Must Read” priority URL list so important pages appear at the top of llms.txt.
  • FEATURE: Background regeneration with progress indicator, plus a separate Preview Diff tool to compare URL changes before regenerating.
  • FEATURE: Live Preview now reflects unsaved form settings.
  • FEATURE: WordPress Site Health checks for file writability, schedule status, and last generation result.
  • FEATURE: Deeper Yoast/Rank Math integration: skip noindex posts and include focus keywords in summaries.
  • FEATURE: Path-level Allow/Disallow rules for custom AI agents.
  • FEATURE: Multilingual language sections when WPML or Polylang is active.
  • FEATURE: Automatic options upgrade merges new settings keys on load (no manual Save required).
  • ENHANCEMENT: DOM-based Markdown converter with improved lists, tables, images, and links for RAG.
  • UI: Refreshed admin interface with updated colour theme, pill tabs, and clearer Tools panel.
  • UI: Improved File Status sidebar alignment and Custom AI Agents layout.
  • BUGFIX: Virtual mode and Live Preview now include the same crawler rules section as physical files.
  • BUGFIX: Analytics no longer records unrecognised browsers as AI bot traffic.
  • BUGFIX: Plugin admin notices (generation status and robots save feedback) display correctly again.
  • BUGFIX: Robots.txt safety validator only blocks a true site-wide User-agent: * + Disallow: / rule, so legitimate per-bot Disallow rules can be saved.
  • BUGFIX: Disabled virtual Markdown/JSONL endpoints return a normal 404 without sending the wrong content headers.
  • BUGFIX: Group-by-type generation now respects the overall item limit across all content types.
  • BUGFIX: Manual regeneration no longer interferes with the separately scheduled regeneration cron.
  • BUGFIX: Author archive links are capped in physical generation to avoid oversized author sections.
  • SECURITY: Additional hardening for admin request handling, REST access, and filesystem writes.

6.3.1

  • FEATURE: Upgraded Markdown Generation to support h4-h6, blockquotes, code blocks, and images.
  • FEATURE: Chart colors are now dynamically generated using HSL (golden angle approximation).
  • SECURITY: Added robust WP_Filesystem API checks in Generator.php and Robots.php.
  • PERFORMANCE: O(1) hash map indexing for AI Analytics charts to drastically improve dashboard speed.
  • PERFORMANCE: Option caching in Admin.php to prevent redundant database queries.
  • PERFORMANCE: Replaced slow HTTP loopback with native WordPress filters to fetch virtual robots.txt rules, preventing deadlock.
  • TWEAK: Clamped virtual_item_limit (500) and get_users queries to prevent memory exhaustion on large sites.
  • TWEAK: Centralised known bots list to Logger::get_known_bots().
  • TWEAK: “Unknown AI Bot” fallback for unrecognised crawlers in Analytics.
  • TWEAK: Translated admin UI tabs using esc_html_e().
  • TWEAK: CSV export now respects the date range selected in the UI.
  • UI: Changed CSS focus styles from outline: none to box-shadow to meet WCAG 2.4.7 accessibility standards.
  • UI: Analytics chart headers update dynamically based on the selected date range.

6.3.0

  • FEATURE: Upgraded the llms.txt generation to achieve 100% compliance with the official standard.
  • TWEAK: URLs are now properly formatted as Markdown lists (- [Title](URL)) to avoid Agentic Browsing errors.
  • TWEAK: Site description is properly output in a blockquote immediately following the H1 header.
  • TWEAK: Links to alternative formats (llms-full.txt and llms.jsonl) have been moved under an ## Optional header.
  • COMPATIBILITY: Verified compatibility with WordPress 7.0.

6.2.2

  • BUGFIX: Fixed an issue in Markdown conversion where elements with HTML attributes (like classes or IDs) were incorrectly stripped instead of being converted.
  • BUGFIX: Added robust error handling and file pointer checks when streaming data to llms-full.txt and llms.jsonl to prevent fatal errors on restricted server environments.
  • BUGFIX: Fixed a settings import bug by replacing WP_Filesystem with standard file reading methods, ensuring imports work on servers that require FTP credentials.
  • BUGFIX: Rewrote IP rate-limiting blocklist to enable full support for IPv6 addresses and CIDR subnets.
  • PERFORMANCE: Transitioned heavy AI analytics queries from non-persistent object caching to Transients, drastically reducing database load on the Analytics dashboard.
  • TWEAK: Restricted the AI Bot Access Logs dashboard widget to administrators only.
  • TWEAK: Added explicit error logging when an XML Sitemap fails to parse to assist in debugging malformed URLs.
  • SECURITY: Addressed a potential SSRF vulnerability in the Sitemap processing logic.
  • TWEAK: Reduced the internal robots.txt loopback timeout to prevent settings page hangs on restricted hosting environments.
  • BUGFIX: Fixed an issue where the CSS cleanup selectors field would always appear blank after saving.
  • BUGFIX: Improved stability when no post types are configured by preventing an unconstrained database query.
  • BUGFIX: Fixed a potential fatal error in the WP-CLI regenerate command when plugin options have not yet been saved.
  • SECURITY: Strengthened the robots.txt safety validator to catch a wider range of dangerous global Disallow configurations.
  • SECURITY: Restricted the filter_date option to a predefined list of allowed values.
  • PERFORMANCE: Increased the analytics data cache lifetime to reduce unnecessary database queries on the admin dashboard.
  • TWEAK: Analytics data is no longer pre-loaded on the admin page when Physical File mode is active.
  • BUGFIX: Fixed several admin field callbacks that could produce PHP notices on fresh installations before settings are first saved.
  • TWEAK: Removed redundant error suppression from the HTML parser; internal error handling is sufficient.
  • MAINTENANCE: Plugin uninstall now cleans up all rate-limiting and block transients from the database.
  • BUGFIX: Fixed an issue where the virtual JSONL endpoint could produce a fatal error when no content types were selected.
  • BUGFIX: Corrected the activation and deactivation routines to ensure they execute reliably in all environments.
  • BUGFIX: Fixed an issue where saved IP blocklist entries were not visible in the settings UI after saving.
  • BUGFIX: Resolved an incompatibility in the robots.txt editor where certain valid characters could be silently removed on save.
  • BUGFIX: Improved error handling in the manual regeneration tool to provide a clearer message when plugin settings have not been configured.
  • TWEAK: Settings changes made via WP-CLI or programmatically now correctly clear caches and update the regeneration schedule.
  • TWEAK: The AI bot activity dashboard widget now appears in all serving modes, with a contextual note when analytics are unavailable.
  • TWEAK: Analytics charts and log tables are no longer rendered on the admin page when bot tracking is inactive.
  • MAINTENANCE: Minor internal code quality improvements and redundant logic removed.
  • BUGFIX: Fixed the Analytics date-range Refresh button. Charts and the detailed bot table were not updating due to mismatched response field names.
  • BUGFIX: Fixed the per-file hits column in the Analytics refresh always showing NaN; the correct database field name is now used.
  • BUGFIX: Removed dead unreachable code after WP_CLI::error() in the CLI regenerate command.
  • BUGFIX: Fixed the Markdown DOM cleanup pass incorrectly emitting wrapper tags (DOCTYPE, <html>, <body>) into the plain-text output, which caused stray heading markers to appear at the start of generated Markdown files.
  • BUGFIX: Removed the unnecessary mb_convert_encoding() HTML-ENTITIES conversion from the Markdown cleaner; UTF-8 is now handled correctly via an explicit charset declaration, eliminating a potential source of character corruption.
  • BUGFIX: Fixed virtual-mode 404 responses for disabled llms-full.txt and llms.jsonl endpoints not terminating correctly after sending the Content-Type: text/plain header, which could allow WordPress to continue template resolution.
  • BUGFIX: Fixed esc_html() being incorrectly applied to robots.txt rule output. As a plain-text file, this would corrupt any rule containing &, <, or > characters into HTML entities.
  • BUGFIX: Fixed the analytics log-pruning cron job silently failing to register on fresh plugin activations. The activation routine runs before the custom cron interval filter fires, so the job now uses the built-in WordPress weekly schedule, which is always available at activation time.
  • BUGFIX: Added missing return statement after the permission check in the simulate-hit AJAX handler for consistency and static analysis correctness.
  • BUGFIX: Corrected the settings export filename date format from Y-d-m (producing 2026-28-04) to the correct ISO 8601 Y-m-d format.
  • BUGFIX: libxml_clear_errors() is now called on the successful XML parse path in the Sitemap parser, preventing libxml error objects from accumulating in memory across recursive sitemap index requests.
  • TWEAK: Aligned the user_agent database write length with the actual varchar(255) column width.
  • MAINTENANCE: Normalised all source file line endings to LF for consistent version control diffs.

6.2.1

  • BUGFIX: Restored missing HTML controls for the Analytics date range selector and per-file breakdown.
  • SECURITY: Fixed missing output escaping for custom crawler rules in the admin UI.
  • COMPATIBILITY: Prefixed core content filters with odyssey_llms_ for improved plugin isolation and linting compliance.
  • BUGFIX: Replaced strip_tags() with the standard wp_strip_all_tags() throughout the codebase.
  • BUGFIX: Improved HTML entity decoding in Markdown exports to prevent malformed character strings.
  • BUGFIX: Removed redundant HTML escaping for titles, taxonomies, and WooCommerce SKUs in plain text formats.
  • ENHANCEMENT: Updated JSON exports to preserve readable unicode characters and slashes.
  • TWEAK: Clarified the WooCommerce product data setting dependency on the Markdown Generator option.

6.2.0

  • SECURITY: Several internal security hardening improvements.
  • BUGFIX: Fixed an issue where certain valid characters were incorrectly stripped when saving robots.txt content.
  • BUGFIX: Fixed an issue where the sitemap tester could reference stale data instead of the live input value.
  • FEATURE: New Site Title and Site Description fields to populate standard header directives in generated files.
  • FEATURE: New Content Ordering option to control how posts are sorted across all generated files.
  • FEATURE: New JSONL Prompt Template field for customising fine-tuning dataset prompts.
  • FEATURE: New Markdown Grouping toggle to organise llms-full.txt output by post type.
  • FEATURE: New Test Sitemap button to validate a configured sitemap URL directly from the settings page.
  • FEATURE: New Custom Bot Rules repeater UI for managing additional AI crawler entries without editing files.
  • FEATURE: REST API endpoints added for integration and developer access.
  • FEATURE: Analytics tab now includes a date range selector and a per-file breakdown table.
  • FEATURE: WP-CLI status command enhanced with scheduling details and a 30-day analytics summary.
  • FEATURE: Automatic analytics log pruning via a scheduled background task.
  • FEATURE: Physical files are removed on uninstall. Analytics data and settings are preserved for clean reinstallation.
  • PERFORMANCE: Various internal refactoring and optimisation improvements.
  • COMPATIBILITY: Improved compatibility with multisite and a broader range of hosting environments.
  • SECURITY FIX: Rate limiter block duration now enforces a minimum of 1 hour, preventing a zero-value option from creating a permanent, unrecoverable IP lockout via a never-expiring transient.
  • BUGFIX: Pruning cron job now correctly uses the built-in WordPress weekly schedule at activation time, ensuring consistent background pruning on all environments.
  • BUGFIX: Robots.txt safety validator now normalises whitespace before checking for dangerous Disallow: / directives, closing a bypass where multiline input could slip through the check.
  • FEATURE: Taxonomy archive URLs are now correctly included in llms.txt when taxonomies are selected in the Content Source settings (previously saved but never written to output).
  • FEATURE: Author archive URLs are now correctly included in llms.txt when the Author Archives option is enabled (previously saved but never written to output).
  • FEATURE: WooCommerce product metadata (Price, SKU, Stock Status) is now correctly appended to llms-full.txt and llms.jsonl when the WooCommerce integration is enabled (previously saved but never applied to output).

Earlier versions

The full changelog for version 6.1.14 and earlier is in the changelog.txt file included with the plugin.