citelayer® – Agent Readiness | MCP Abilities, Schema.org

Changelog

1.13.0

• NEW: Two writing citelayer abilities for the WordPress Abilities API — “citelayer/run-readiness-scan” runs the AI readiness checks and stores the result, “citelayer/set-llms-txt-exclusion” includes a single post in llms.txt or holds it back.
• NEW: Every ability now carries an access level (read or write) and runs through a shared execution wrapper that checks the administrator capability a second time, so the permission callback is no longer the only barrier.
• NEW: Toggle “Read-only mode” on the WebMCP settings page, on by default, so neither a new install nor an update makes a site writable for an AI agent without a deliberate decision. While it is on, the writing abilities are not registered at all and a call that still arrives is refused; the reading abilities are unaffected.
• NEW: /.well-known/mcp/context.json now reports the read-only state and the access level of every registered ability.
• IMPROVEMENT: An unexpected error inside an ability is returned as a WP_Error instead of escaping as an exception, so a single failure no longer empties an MCP client’s tool list.
• IMPROVEMENT: The readiness scan started from an ability stops at a runtime budget instead of running into the PHP time limit, and reuses a recent result unless “force” is set. A scan that stops early keeps the previous scan timestamp, so the dashboard does not show a partial run as a fresh one and the next call can continue right away.
• IMPROVEMENT: A completed ability scan now replaces the stored result instead of merging into it, so checks that no longer exist — for example the WooCommerce ones after deactivating WooCommerce — no longer linger in the score.
• IMPROVEMENT: The citelayer ability category is described as “AI visibility data and controls”, since it is no longer read-only.
• REMOVED: SEOPress is no longer detected or handled specially. citelayer treats it like any other SEO plugin it does not know. Four things change on a site with SEOPress active: citelayer’s own Schema.org markup stays on and is emitted next to SEOPress’s own markup, so the front page can carry the same entry twice; a post you marked noindex in SEOPress is no longer excluded from llms.txt and gets a Markdown version again; citelayer emits its own article:modified_time meta tag again, which appears twice if you switched on SEOPress’s “Display date in Google search results” option (it is off by default); and, if you never picked a Markdown scope yourself, citelayer now follows its own selection of post types instead of SEOPress’s sitemap, which can change which pages are served as Markdown. Your saved settings are untouched — only defaults that were derived from SEOPress being present. If you rely on SEOPress and want this back, please open an issue.
• FIX: If an SEO plugin was installed but not set up — Rank Math before its setup wizard has run, or All in One SEO with schema markup switched off — citelayer held back its own Schema.org output while the SEO plugin produced none either, leaving the page with no structured data at all. citelayer now checks whether the SEO plugin actually outputs markup on the current request and steps in when it does not.
• IMPROVEMENT: The Schema.org status in the compatibility settings and in the abilities now reads the same decision that drives the front end, instead of a second set of rules next to it, and it no longer changes depending on which page the request came from. The schema status ability additionally reports how far the SEO plugin takes over.
• IMPROVEMENT: New plugin title “Agent Readiness | MCP Abilities, Schema.org” and a new short description, in the plugin directory and in the plugin list inside WordPress. In the tag list “schema” is replaced by “mcp” and “llms-txt” by “ai-agents”. The plugin slug, your settings and your data are unchanged.
• IMPROVEMENT: The plugin description now leads with the MCP abilities and with the diagnosis that tells you when an endpoint stops working, and describes Schema.org as the supporting part it is.
• IMPROVEMENT: llms.txt is no longer presented up front. The feature stays exactly as it is, but the description now names the studies: across 300,000 domains there is no statistically significant difference in AI citation frequency between sites with and without the file. Nothing in your setup changes.
• IMPROVEMENT: The AI Readiness Score no longer counts the llms.txt check, so a site that gets everything measurable right is no longer pulled down by a file the AI crawlers demonstrably ignore. The check itself stays, still shows its status, and says in the result why it is not counted. Expect your score to go up if llms.txt was one of your red checks.

1.12.0

• NEW: Seven read-only citelayer abilities for the WordPress Abilities API (WordPress 6.9+), so MCP clients can read the AI readiness score, llms.txt status, bot analytics, AI blind spots, AI referral stats, Schema.org status and compatibility status.
• NEW: Toggle “Enable citelayer Abilities” on the WebMCP settings page, on by default.
• NEW: /.well-known/mcp/context.json reports the abilities status (available, disabled, wp_too_old) together with the registered ability names.
• FIX: /.well-known/mcp/context.json now sends a one-hour HTTP cache lifetime instead of seven days, so client, CDN and agent caches pick up capability changes shortly after you make them.
• FIX: The AI referral counters are now cached per reporting window, so asking for 7, 30 or 90 days no longer returns the numbers of the window you asked for before.
• FIX: Switching a citelayer feature on or off for the first time now refreshes /.well-known/mcp/context.json right away instead of leaving the previous capability list in place.
• FIX: The Schema.org status no longer claims WebSite and Article output while an SEO plugin owns the schema markup; it reports the site identity citelayer merges into that plugin instead.
• FIX: Abilities without parameters now publish a valid JSON Schema, so strict MCP clients accept them.
• COMPATIBILITY: The citelayer ability category is only registered when no other plugin registered the same slug first.
• COMPATIBILITY: On WordPress older than 6.9 no ability is registered and no notice is raised. Requires at least stays 6.4.

1.11.3

• FIX: Schema caches separate language and site URL contexts, preventing mixed-language identity IDs and translated article data.
• FIX: Organization, Person, WebSite and Article output filters apply on cached reads, preserving current site-owned entity mappings.
• FIX: Schema cache invalidation covers all language variants, including persistent object caches. Existing cache entries are bypassed after the update.

1.11.2

• FIX: Markdown responses preserve headings, links, lists, emphasis, images, blockquotes, and code by loading the HTML-to-Markdown library’s standard converters and defaults.
• FIX: HTML5 elements no longer trigger libxml warnings during Markdown conversion.
• FIX: Markdown caches regenerate automatically after plugin updates, including previously flattened responses.
• COMPATIBILITY: Tested with WordPress 7.1.

1.11.1

• FIX: llms.txt diagnostics now recognize Rank Math Enhance-mode output that contains the Citelayer enhancement marker, so it is no longer misreported as a physical shadow file.
• FIX: Endpoint diagnostics now reserve “Physical file detected” for confirmed root llms.txt/llms-full.txt files and use neutral wording when another generator or server rule serves the public endpoint.

1.11.0

• NEW: Compatibility mode now detects Rank Math AI Visibility and explains how it complements Citelayer’s AI-readable site infrastructure.
• NEW: WebMCP context now exposes safe Rank Math AI Visibility integration metadata when available, without copying Rank Math scores, brands, competitors, or query transcripts.
• IMPROVEMENT: Compatibility wording now distinguishes external AI brand monitoring from local bot/referral evidence, llms.txt, Markdown, Schema.org, WebMCP, and UCP delivery signals.

1.10.2

• FIX: Bot Analytics now exposes a local Claude, Perplexity, and Gemini self-check for bot User-Agent, referrer, and UTM matching.
• IMPROVEMENT: AI Referrals now keeps a privacy-minimal recent signal diagnostic log for accepted, rejected, and deduplicated endpoint attempts.
• IMPROVEMENT: AI Referrals and Crawler Logs now explain repeated AI-search behavior more explicitly, including cached/indexed answers and missing browser referrers.

1.10.1

• FIX: AI Referrals now resolves known AI sources server-side from referrer and UTM signals, with stricter host matching and expanded source coverage.
• IMPROVEMENT: Bot Analytics now explains crawler activity, external AI visibility/citations, and human AI referrals as separate signals.
• IMPROVEMENT: AI Referrals empty state now shows tracking diagnostics and practical checks for REST, frontend script, referrer/UTM, cache/consent/CSP blockers.

1.10.0

• NEW: Schema settings now explain what each Schema Types toggle controls, when to keep the recommended defaults, and how Enhance Mode treats existing SEO plugin graph data.
• NEW: Organization Schema type selection is now hierarchical, showing more specific Schema.org subtypes only when they apply and saving the final selected subtype for JSON-LD output.
• IMPROVEMENT: Local-business-specific Schema fields now appear dynamically for location-based Organization types.

1.9.0

• NEW: AI Readiness Scanner adds Brand Demand Signals guidance, checking local brand entity/profile readiness while clearly marking real brand search volume as an external audit signal.
• IMPROVEMENT: Scanner results can link administrators to the AI Visibility Audit for measured brand-search-demand analysis.

1.8.0

• NEW: AI Readiness Scanner adds a Topic Coverage check for distinct H2/H3 subtopics and internal topic-cluster links.
• NEW: Organization sameAs settings now provide dedicated YouTube and Reddit profile fields and prioritize them before additional profiles.
• IMPROVEMENT: Markdown settings now highlight deep content discovery so evergreen guides, documentation, and other useful pages stay available to AI agents.

1.7.2

• FIX: WebMCP now exposes UCP as discovery metadata only, so WooCommerce presence no longer implies product catalog, cart, checkout, or Full UCP support.
• FIX: UCP Discovery now publishes explicit capability status and documented filters for active, licensed commerce add-ons.
• IMPROVEMENT: llms.txt discovery links now describe UCP/WebMCP as capability-status metadata instead of shipped autonomous shopping support.

1.7.1

• FIX: Premium/UCP wording now distinguishes shipped discovery and controlled cart/checkout handoff from future Full UCP support.
• FIX: UCP discovery upgrade messaging no longer claims Full UCP support before the Woo add-on milestone is complete.

1.7.0

• NEW: Schema settings now support explicit site identity selection between Organization and Person, with valid Schema.org Organization subtypes and person role presets.
• NEW: Organization/Person identity fields, social profiles, contact details, address, local-business geo/opening-hours, expertise, and JSON-LD preview are editable directly in the Citelayer admin.
• IMPROVEMENT: Plain Yoast SEO and Yoast SEO Premium no longer disable Citelayer identity schema fields; overlapping local identity fields are delegated only when Yoast Local SEO is active.
• FIX: Rank Math, Yoast, and All in One SEO schema enhancement now merges the active site identity into existing Organization, LocalBusiness, or Person graph nodes without duplicate identities.
• FIX: Cached SEO plugin detection is validated against currently active plugins so a recently disabled Yoast Local SEO add-on cannot keep Citelayer in local-identity defer mode.

1.6.1

• FIX: Yoast llms.txt saves are now re-enhanced automatically in Compatibility Enhance mode, keeping Citelayer additions visible after Yoast regenerates its physical llms.txt file.
• FIX: llms.txt diagnostics no longer classify an intentionally enhanced Yoast static file as stale shadow output when it contains the Citelayer footer.

1.6.0

• NEW: llms.txt admin diagnostics now include a manual “Check again” action that clears stale endpoint status and re-runs live public/fallback checks.
• NEW: Admins can download a backup of physical root llms.txt/llms-full.txt shadow files and remove those exact files safely from the llms.txt screen.
• IMPROVEMENT: Endpoint diagnostics now expose machine-readable reason codes and show clearer copy for static-file shadowing, missing UTF-8 charset, and blocked public routes.

1.5.9

• FIX: llms.txt diagnostics now detect stale static or SEO-generated root files that shadow Citelayer output, compare public endpoints with the Citelayer fallback generator, and show Content-Type, charset, and Citelayer footer details.
• FIX: Replace-mode remediation now clears both root llms.txt and llms-full.txt shadow files, and /.well-known/llms.txt routes to the Citelayer generator instead of falling through to the homepage.

1.5.8

• FIX: Compatibility Mode notices can now be dismissed reliably on the WordPress Plugins screen. The notice CSS and JavaScript are loaded before notices render, and dismissals persist per admin user.

1.5.7

• FIX: AI Readiness Scanner now distinguishes a blocked public /llms.txt URL from a broken generator when the Citelayer fallback endpoint still returns valid content.
• IMPROVEMENT: llms.txt admin status now checks both /llms.txt and /llms-full.txt, shows canonical and fallback HTTP results, and points admins to server/CDN/WAF rules when public .txt paths are blocked before WordPress.
• CHORE: PHPUnit composer script now runs with the documented 512M memory limit so the committed WordPress and WooCommerce stubs load reliably.

1.5.6

• IMPROVEMENT: Tested and compatible with WordPress 7.0, with a full WordPress Plugin Check compliance pass (coding standards, input sanitization, database query safety).
• FIX: No longer outputs a duplicate Organization schema node when running alongside Rank Math, Yoast SEO, or All in One SEO — citelayer now merges its data into the active SEO plugin’s schema graph instead of adding a second node.
• FIX: Markdown (.md) endpoint now resolves all public content — posts, custom post types, and products — not just pages.
• IMPROVEMENT: WooCommerce and Easy Digital Downloads functional pages (cart, checkout, account) are now excluded from llms.txt and Markdown output, reducing noise for AI crawlers.
• FIX: WebMCP context endpoint (/.well-known/mcp/context.json) now returns the correct application/json content type so strict AI/MCP clients accept the response.
• FIX: llms.txt discovery links (UCP, WebMCP) now point to the correct registered endpoints.
• FIX: Markdown output is now restricted to published content only (defense-in-depth hardening).

1.5.5

• FIX: WooCommerce product Markdown export no longer fails in background tasks or contexts where the product is referenced by ID rather than loaded as an object.

1.5.4

• IMPROVEMENT: Schema Types settings now show blur overlay with direct link to SEO plugin when in Enhancement Mode
• NEW: Product Feed now supports Easy Digital Downloads (simple + variable pricing)
• FIX: EDD addon Settings button on license page linked to non-existent page
• NEW: Content-Signal HTTP header now sent on all frontend pages (not just Markdown endpoint)

1.5.3

• FIX: WooCommerce hidden products (visibility=hidden) now excluded from llms.txt
• FIX: WooCommerce hidden products now excluded from product feed XML

1.5.2

• FIX: llms.txt cache not invalidated when settings are changed (exclude URLs, description, title, etc.)
• FIX: Wildcard URL exclude patterns (e.g. /shop/*) never matched due to preg_quote escape bug
• FIX: Plain path exclude patterns (e.g. /warenkorb) only checked inside http-URL branch due to missing brace

1.5.1

• IMPROVEMENT: AI Score expanded from 7 to 11 checks — 4 new content-quality checks
• NEW: Opening Summary check (first paragraph 30–80 words)
• NEW: Heading Structure check (≥2 H2 + ≥1 H3)
• NEW: Internal Linking check (≥2 internal links)
• NEW: FAQ Section detection (FAQ blocks, headings, and Q&A patterns)
• Rebalanced check weights for better signal distribution

1.5.0

• NEW: Per-Page AI Visibility Score — editor metabox shows 0–100 score with 7 AI readiness checks
• NEW: Content Freshness Signals — Last-Modified header + article:modified_time OG meta (SEO-plugin aware)
• IMPROVEMENT: Scanner now differentiates AI Training Bots (blocking = good) from AI Search Bots (blocking = bad)
• IMPROVEMENT: Bot list expanded to 86 bots (+PhindBot, img2dataset, KangarooBot, Sidetrade, webz.io, Bingbot-AI)
• IMPROVEMENT: Bot Analytics dashboard shows category badges (AI Training, AI Search, AI Agent, AI Enterprise)

1.4.3

• IMPROVEMENT: Keyword-optimized plugin display name for better wp.org discoverability
• Updated plugin name to “citelayer® – AI SEO & Visibility | llms.txt, Bot Analytics, Schema.org”

1.4.2

• FIX: Remaining PHPCS violations — file-level disables for template views and DB query files
• FIX: Function call indentation in Compatibility.php
• FIX: Inline phpcs:ignore for long translation strings

1.4.1

• FIX: Resolved all PHPCS WordPress Coding Standards violations
• FIX: Short ternary operators replaced with explicit ternaries (ProductFeed, ProductFeedCheck, LLMS_TXT)
• FIX: Yoda conditions, indentation, embedded PHP tag formatting
• FIX: Trailing whitespace and line length violations
• IMPROVEMENT: Added phpcs:disable/enable blocks for mixed PHP/HTML template methods

1.4.0

• NEW: AI Referral Tracking — track visitors arriving from ChatGPT, Perplexity, Claude, Gemini, Copilot, You.com, Phind, Kagi, and Andi
• NEW: “AI Referrals” dashboard tab with stat cards (Visits from AI, Top Source, Trend %), trend chart, top sources breakdown, and paginated referral log
• NEW: Lightweight frontend tracking via sendBeacon — zero impact on page load performance
• NEW: Rate-limited REST endpoint with origin validation and IP hashing (GDPR/DSGVO compliant)
• NEW: llms.txt now includes Markdown URLs for each page, helping LLMs discover optimized content versions
• NEW: 90-day auto-cleanup for referral data (same as bot analytics)
• IMPROVEMENT: SVG trend chart now supports custom colors (reusable across tabs)
• 100% local — all referral data stored in your WordPress database, no external services

1.3.5

• NEW: Bot Analytics — DB-backed AI bot visit tracking with real-time dashboard
• NEW: Expanded bot database from 17 to 62 AI/LLM bots (OpenAI, Anthropic, Google, Meta, xAI, DeepSeek, Mistral, Cohere, HuggingFace, and more)
• NEW: Bot categories (Confirmed AI, AI Assistant, AI Search, AI Enterprise, Possible AI)
• NEW: Admin dashboard with stat cards (7d/30d/90d), SVG trend chart, and top bots visualization
• NEW: AI Blind Spots — discover pages invisible to AI bots with AI Discovery Score (0-100)
• NEW: CSV export for bot visit logs with date/bot/path filtering
• NEW: Cache plugin detection (WP Rocket, LiteSpeed, W3TC, WP Super Cache) with setup guides
• NEW: 90-day auto-cleanup to prevent database bloat
• FIX: Settings link on plugins page now points to correct dashboard URL
• IMPROVEMENT: Clean uninstall — removes DB table, options, and transients on plugin deletion
• 100% local — no external accounts or API calls needed

1.3.4

• FIX: AI Crawler Detection now works in all compatibility modes (including Rank Math enhance mode)
• FIX: Per-post exclude meta box only shown when citelayer generates llms.txt
• PERF: Batch noindex queries (N per-post queries → 3-4 bulk queries)
• PERF: Capped posts at 500/type and products at 1000 (filterable)
• PERF: Reduced Scanner HTTP timeouts from 10-15s to 5s

1.3.3

• NEW: AI Crawler Detection — see which AI bots visit your llms.txt (GPTBot, ClaudeBot, PerplexityBot, and 14 more)
• NEW: in HTML head for AI discoverability
• NEW: Per-post exclude toggle — exclude individual pages from llms.txt via meta box
• NEW: Custom title and description overrides for llms.txt header
• NEW: Max words cap for llms-full.txt (default 500, configurable 50–10.000)
• NEW: Taxonomy inclusion in llms-full.txt (categories, tags, custom taxonomies)
• NEW: Meta inclusion in llms-full.txt (published/modified dates, URL, WooCommerce SKU + price)
• Privacy-first: All data stored locally, no external server calls

1.3.2

• NEW: llms-full.txt — full content export for deep AI understanding
• NEW: Configurable post types for llms.txt (pages, posts, products, CPTs)
• NEW: SEO noindex respect (Yoast, Rank Math, SEOPress, AIOSEO)
• NEW: URL exclude patterns with wildcard support
• Improved: Post type grouping with section headers
• Improved: Proper > description blockquote per llmstxt.org spec
• Improved: Admin page with post type selection and pattern examples

1.3.1

• NEW: XML Product Feed Generator — OpenAI Commerce Protocol compatible
• Serves product feed at /product-feed.xml
• Supports simple, variable, and external products with variations
• GTIN/EAN auto-detection from common meta fields
• WP-Cron scheduled regeneration (hourly/daily/weekly)
• Transient-based caching with configurable TTL
• New admin page: citelayer → Product Feed
• Scanner: added Product Feed validation check

1.3.0

• NEW: AI Readiness Scanner — comprehensive site audit for AI visibility
• 7 automated checks: Schema.org, robots.txt (8 AI bots + Cloudflare), llms.txt, Content-Signal headers
• WooCommerce checks: Product descriptions, images, stock status
• Visual readiness score dashboard (0-100)
• Async scanning via REST API
• New admin page: citelayer → AI Scanner

1.2.2

• Rewritten plugin description — user-centric, benefit-driven (AI Visibility, GEO, AEO)
• Updated FAQ in plain language
• Optimized tags for WordPress.org discoverability
• Updated plugin banner images

1.2.1

• Fixed WordPress.org deploy trigger (SVN deploy now runs in same workflow as auto-tag)
• Added manual deploy fallback with dry-run support

1.2.0

• Added automated WordPress.org SVN deploy workflow (deploy on GitHub Release)
• Added plugin directory assets (icons, banners)
• Added .distignore for clean SVN deployments
• Added readme/assets update workflow (update WordPress.org without new release)
• Streamlined auto-tag workflow (ZIP build now handled by deploy action)

1.1.9

• Re-included composer.json in distribution ZIP (required when vendor/ is present)

1.1.8

• Removed vendor CLI binaries (bin/html-to-markdown) from release ZIP
• Excluded composer.json from distribution
• Improved forbidden file detection in build workflow

1.1.7

• Lowered PHP requirement from 8.1 to 7.4 (code + vendor deps confirmed compatible)
• Fixed deploy workflow: .gitattributes and README.md now excluded from production
• Added PHP 7.4 to CI syntax check matrix

1.1.6

• Converted inline scripts to wp_add_inline_script() for WP Codex compliance
• Extracted inline styles to external CSS files with wp_enqueue_style()
• Updated contributor and Author URI metadata

1.1.5

• Fix fatal error: rebuild vendor with composer –no-dev for release ZIP

1.1.4

• Exclude dev dependencies from release ZIP (phpstan, phpcs, etc.)

1.1.3

• Plugin name fix

1.1.1

• Respecting WP / WordPress trade mark

1.1.0

• Code Quality: PHPCS + PHPStan Audit — alle Violations behoben (PHPCS 260→1, PHPStan 649→0)
• Tooling: phpcs.xml.dist, phpstan.neon, Composer Scripts (lint, fix, analyse)
• CI: Strikte PHPCS- und PHPStan-Jobs (kein continue-on-error mehr)
• Infrastructure: Pre-Commit-Hook für automatisches Code-Formatting
• Fix: Schema.php — post_author Type-Cast für get_the_author_meta()

1.0.0

• Production release
• Version synchronization (header, constant, readme.txt)
• Security: Admin email no longer exposed in llms.txt by default
• UCP Discovery: Dynamic platform detection (WordPress vs WooCommerce)
• WebMCP: Removed non-implemented endpoint references
• WebMCP: Fixed option check consistency
• Added PHPCS configuration (WordPress Coding Standards)
• Added PHPStan Level 6 configuration
• Comprehensive CI pipeline (syntax, PHPCS, PHPStan, integrity checks)
• Removed development artifacts from repository

0.3.1

• Compatibility V3.1: Multi-Family Safeguard — Conflict State detection
• Admin: Conflict State banner on all admin pages
• llms.txt: 503 response in Conflict State
• Schema: No output in Conflict State
• WebMCP context.json discovery endpoint

0.3.0

• Compatibility Layer V3: Enhance-Fix
• serve_llms_txt() uses should_register_own_route() to prevent double output
• Yoast Fallback: Citelayer takes over when Yoast has no physical file
• State-Reset on plugin deactivation
• Improved AIOSEO detection (defined + class_exists checks)

0.2.0

• Composer Support: league/html-to-markdown integration
• PSR-4 Autoloading with SPL fallback
• Feature-Detection on module level (Rank Math llms.txt, Yoast llms.txt)
• All classes refactored to Citelayer\Plugin namespace
• File structure: includes/ → src/

0.1.0

• Initial release
• llms.txt generation with WooCommerce awareness
• Schema.org JSON-LD (Organization, WebSite, Article, Product)
• UCP Discovery endpoint
• Markdown for Agents with Content Negotiation
• Content-Signal header configuration
• Settings page in WordPress admin