Ligase WP Registry grade A. No security issues foundWP RegistryNo security issues foundOpen the reportA

Schema.org JSON-LD for WordPress — entity graph, AI-citation optimization, WooCommerce-aware merchant listings, and a single-source-of-truth field contract.

What is Ligase?

Ligase generates a single consolidated @graph of schema.org JSON-LD for every page of a WordPress site — connecting BlogPosting/Product/Recipe/JobPosting nodes to a unified Person (author) and Organization (publisher / OnlineStore) through @id references. The graph is rendered server-side in wp_head so it's visible to AI crawlers (GPTBot, ClaudeBot, PerplexityBot) that don't execute JavaScript.

For: professional bloggers, news publishers, WooCommerce stores, recipe sites, job boards, and forums that want Google rich results and AI-citation visibility without writing schema by hand.

What's new in 2.4.x

• PodcastSeries schema (2.4.19) — landing-page hub markup for podcast websites. Per-page opt-in with sameAs (Spotify / Apple Podcasts / YouTube), webFeed (RSS), numberOfEpisodes. Author resolved through the shared BlogPosting::author_ref_id() so org_author_mode + ligase_is_redakcja decisions apply.
• Person extension — personal-brand SEO pack (2.4.18-19): five new repeater fields per author profile —
  • worksFor external (override Organization @id ref with your own firm)
  • affiliation (loose ties — advisory, partnerships)
  • subjectOf (cross-link to external interviews / write-ups where you're the topic)
  • workExperience → emitted as schema.org-canonical role-property pattern inside worksFor array
  • award (recognition with optional issuer + year)
  • agentInteractionStatistic (InteractionCounter — YouTube subs, Spotify plays, LinkedIn followers — AI/LLM authority signal)
• Smart schema-type detection in admin (2.4.10) — Pages auto-classify by slug + title heuristics: Kontakt → ContactPage, O nas → AboutPage, Koszyk → CheckoutPage, Sklep → CollectionPage, FAQ → FAQPage, plus heuristics for blog hubs.
• Google open-web popularity badges (2.4.20) — meta-box shows next to every schema type a color-coded badge with adoption tier (10M+, 1M-10M, 100K-1M, 10K-100K, 1K-10K) sourced from schemaorg/schemaorg public stats. Helps users decide whether enabling a niche type makes sense for their site.
• Schema.org Validator compliance (2.4.22) — worksFor array now uses canonical role-property pattern (OrganizationRole with nested worksFor, not non-standard workExperience). Plus refundType + returnShippingFeesAmount on MerchantReturnPolicy, @type stamping for handlingTime/transitTime (QuantitativeValue with unitCode: DAY), and image + address on Organization/OnlineStore for "Firmy lokalne" / Local Business Google check.
• Output-buffer scrubbing (2.4.13-14) — when standalone_mode is on, an ob_start callback strips foreign BreadcrumbList / Article / Product JSON-LD scripts injected by themes (XStore, Flatsome, Woodmart) that filter hooks can't catch. Idempotent: leaves Ligase's own @graph block untouched.
• Multi-agent code audit fixes (2.4.8-9) — 30+ production bugs fixed: score-killer typos (organization_name → org_name across 7 places), shippingDetails removed from OnlineStore (invalid in schema.org), URL-template encoding fix for SearchAction.target (Google's Sitelinks Search Box), JobPosting country code validation, VideoObject empty-field gate, has_published_posts deprecated-API fix, secret redaction in settings export, suppressor static-state leak, Polish/EU number formatting ("1 299,90 zł" → 1299.9).
• OPcache auto-reset (2.4.10) — fires on plugin activation and upgrader_process_complete. Eliminates the regression class where shared-host PHP-FPM kept the old compiled sanitize() whitelist, causing newly added checkboxes to silently revert on Save.
• Tabbed settings persistence (2.4.14) — hidden-input pattern + array_key_exists() in sanitize() so saving one tab no longer wipes checkboxes in other tabs.
• Audytor @graph unwrap (2.4.17) — auditor now reads each @graph node separately and picks the relevant one by post type, instead of scoring the envelope (which produced 0/100 across all posts).

Full release notes: readme.txt. Architectural history: docs/audit-history/. Data-driven coverage analysis: docs/google-stats-coverage-2026-05.md.

Schema types (v2.4.22 — 25 types)

Publishing & authorship

E-commerce

Editorial & community

Field-contract system

A single declarative file (includes/class-field-contract.php) defines, per @type:

'Product' => [
    'fields' => [
        'name' => [
            'level'    => 'required',
            'sources'  => [ 'manual:', 'wc:name', 'post:title' ],
            'sanitize' => 'text',
        ],
        'offers.price' => [
            'level'    => 'required',
            'sources'  => [ 'manual:', 'wc:price' ],
            'sanitize' => 'float',
        ],
        // …
    ],
],

Ligase_Field_Resolver walks each contract for a given (type, post_id), tries each source in order, sanitizes, and assembles a nested JSON-LD node. It reports:

• per-field state: auto / manual / missing_required / missing_optional
• overall eligible — false when any required field is empty
• list of missing_required keys

The eligibility gate stops Ligase from emitting half-formed merchant listings (e.g. Product without Offer is downgraded to a snippet rather than fabricated). Manual overrides are persisted in _ligase_override[Type][key] post meta and always win over auto sources; clearing the input reverts to auto. Auto values are never persisted — they're computed at render time on every cache miss, so stale prices/dates are impossible.

The in-editor readiness panel (sidebar metabox) calls Ligase_Readiness::for_post() via AJAX and shows the user exactly which fields are filled, by which source, and which required fields are still missing — so non-technical editors can fix rich-result issues without leaving the editor.

AI Search Readiness Score (0–100)

Site-wide and per-post score measuring how ready content is to be cited by AI engines (Google AI Overviews, ChatGPT, Perplexity). Checks include entity-graph linking (@id references), Wikidata sameAs density, image dimensions ≥ 1200 px, author completeness (knowsAbout, jobTitle, sameAs), dateModified discipline, presence of about/mentions with Wikidata, NER LLM coverage. Pipeline-aware: the score reads _ligase_wikidata_suggestions, _ligase_ner_api_results, _ligase_about_entities so it grades real AI signals, not vanity field-presence checks.

Schema Auditor

Detects and (optionally) suppresses or supplements competitor SEO plugins' schema:

• Yoast SEO, Rank Math, All in One SEO, SEOPress, The SEO Framework, Slim SEO, The Events Calendar.
• Three modes: Scan (report only), Supplement (additive — apply_supplement()), Replace (full takeover below score threshold + restore_replacement() for undo).
• Type-aware scoring rubrics: separate scoring for Article, Event, Product, LocalBusiness, Recipe, FAQ, HowTo, Video, Organization, Person — Event no longer scored against an Article rubric and force-replaced.

Entity-detection pipeline

Level 1: WordPress Native       (tags, categories, author)         ~0ms
Level 2: Structural             (Wikipedia links, YouTube, blocks)  ~5ms
Level 3a: Regex NER             (Polish lemmatization-aware dedup) ~20ms
Level 3b: LLM NER (cron)        (OpenAI / Anthropic / Google /
                                Dandelion — merged with regex)     async
Level 4: Wikidata lookup        (locale → en fallback, exact-label
                                match guard, single-confidence)    async

LLM results take precedence on entity-name conflicts; Wikidata sameAs attaches only when the label exactly matches the entity name (avoids wrong-entity linking).

More features

• Google Search Console — rich-results dashboard (clicks, impressions, CTR per schema type).
• Gutenberg sidebar — live schema preview + validator.
• Import from Yoast / Rank Math / AIOSEO — handles attachment-ID logos, AIOSEO v3 serialized PHP, v4.5+ nested social URLs.
• WPML / Polylang — inLanguage BCP47 + sameAs sync across translations.
• Weekly Health Report — email digest of schema issues.
• Speakable schema — configurable CSS selectors per post.
• Auto-repair — ISO 8601 date fixes, headline truncation, type conversion.
• WooCommerce cache invalidation — hooks into 5 product/stock events; price changes never lag the schema cache.
• i18n-ready — wp_set_script_translations + wp-i18n for JS strings (translation source needs migration to English for wordpress.org; current source is Polish).

Requirements

• WordPress 6.0+
• PHP 8.0+
• No external runtime dependencies (ext-json, ext-mbstring standard)
• Optional: WooCommerce 6.0+ for store features
• Optional: PHP API key for one of OpenAI / Anthropic / Google NLP / Dandelion to enable LLM NER

Installation

From ZIP

1. Download the latest release from Releases.
2. WordPress → Plugins → Add New → Upload Plugin → select the zip → Install Now → Activate.

Post-installation setup

1. Ligase → Settings → fill organization data (name, logo URL ≥ 112×112 square, email, Wikidata ID).
2. Add social media links (Wikidata + LinkedIn = strongest entity signals).
3. Edit author profiles — jobTitle, knowsAbout, sameAs.
4. (Stores) Settings → Store → return + shipping country, return days, shipping rates.
5. Check Ligase → Dashboard for AI Search Readiness Score.

Developer API

Filters

// Modify the full @graph
add_filter( 'ligase_schema_graph', function ( array $graph ): array {
    $graph[] = [ '@type' => 'Event', 'name' => 'My Event' ];
    return $graph;
} );

// Modify a single type node
add_filter( 'ligase_blogposting', function ( array $schema, int $post_id ): array {
    $schema['speakable'] = [
        '@type'       => 'SpeakableSpecification',
        'cssSelector' => [ '.entry-summary' ],
    ];
    return $schema;
}, 10, 2 );

// Extend the field contract from a theme/plugin
add_filter( 'ligase_field_contract', function ( array $contract, string $type ): array {
    if ( $type === 'Product' ) {
        $contract['fields']['offers.priceValidUntil']['level'] = 'required';
    }
    return $contract;
}, 10, 2 );

// Which schema types appear in the readiness panel for this post
add_filter( 'ligase_readiness_panel_types', function ( array $types, int $post_id ): array {
    if ( has_category( 'news', $post_id ) ) {
        $types[] = 'NewsArticle';
    }
    return $types;
}, 10, 2 );

Available type filters: ligase_blogposting, ligase_person, ligase_organization, ligase_website, ligase_breadcrumb, ligase_product, ligase_productgroup, ligase_recipe, ligase_jobposting, ligase_discussionforumposting, ligase_podcastseries, ligase_service, ligase_event, ligase_faqpage, ligase_howto, ligase_review, ligase_localbusiness.

WP-CLI helper

# Print readiness report for a post
wp eval 'var_export( ligase_readiness( 123 ) );'

Testing & quality gates

composer install
composer test            # PHPUnit
composer phpcs           # WordPress Coding Standards
composer phpstan         # PHPStan level 5

GitHub Actions runs all of the above on every push, plus a smoke test that boots WordPress, activates the plugin, creates a post, and verifies that `` inside the JSON is properly escaped (regression guard for the XSS fix).

License

GNU General Public License v2.0 or later.

Author

Built by Marcin Żmuda · Report a bug · Releases