Upgrading to 1.4.0

This guide covers upgrading from v1.3.x to v1.4.0. This release adds AI-first discovery surfaces (llms.txt, IndexNow, AI-crawler robots controls, RSS/Atom feeds, OG image generator), promotes the AI Feature Suite to require artisanpack-ui/ai: ^1.2, and hardens the schema graph, feed generator, and analyzers uncovered by review of the prior betas.

Summary

• New: llms.txt AI-discovery manifest generator (#68).
• New: IndexNow submitter + sitemap-ping path (#69).
• New: Explicit AI-crawler robots controls, including a working default_allow kill-switch (#70, hardened here).
• New: Focus-keyword + first-paragraph AI-readiness analyzers (#71).
• New: RSS 2.0 and Atom 1.0 feed generator (#72).
• New: Organization sameAs + logo (with width/height support) on the schema graph (#73).
• New: OG-image generator service — MightyShare-style social cards rendered via GD, deterministically cached, alpha-preserving (#74).
• New: LocalBusiness dated / holiday OpeningHoursSpecification support (#75).
• New: apSeoAddSchema() helper + SchemaCollector service (#77).
• New: SchemaService render-pipeline centralization (#78).
• New: Every graph node emits a stable @id; Article.publisher now references the Organization by @id so Search Console links the entities across the graph.
• New: Route toggles for llms.txt, /feed.xml, /feed.atom, and the IndexNow key file (seo.llms_txt.route_enabled, seo.feeds.route_enabled, seo.indexnow.route_enabled).
• New: seo.feeds config block (title, description, per_page, cache_ttl, feed_id for Atom).
• Changed: artisanpack-ui/ai constraint bumped to ^1.2. The SEO AI Livewire components now compose the ChecksFeatureToggle and InteractsWithAiFeature traits (#76 / #92). Downstream components that extend the SEO AI components must not redeclare public bool $isLoading or public ?string $error — the trait declares them and hydration throws a TypeError on type conflicts. See "Breaking (soft)" below.
• Fixed: default_allow=false on the AI-crawler controls now actually blocks every group unless blocked is explicitly false; previously a silent no-op.
• Fixed: RSS/Atom output now strips XML 1.0 forbidden control characters and rejects javascript: / data: link schemes so feed validators no longer choke and feed readers no longer render stored-XSS URLs.
• Fixed: LlmsTxtGenerator collapses newlines in titles and escapes [ / ] in descriptions so a Markdown list item can no longer shatter mid-line.
• Fixed: IndexNow surfaces per-URL rejections carried in an otherwise-200 response body (e.g. UnverifiedHost) as success=false.
• Fixed: FocusKeywordAnalyzer and AiReadinessAnalyzer now handle non-ASCII text correctly (multibyte case folding, non-Latin word counting).

Update the Package

composer update artisanpack-ui/seo

If you use the AI Feature Suite, ensure your project is on PHP 8.3+ and Laravel 12+ before upgrading:

composer require artisanpack-ui/ai:^1.2 --dev

Breaking (soft): AI Livewire trait refactor

ChecksFeatureToggle + InteractsWithAiFeature now supply the public bool $isLoading and public ?string $error properties on every SEO AI Livewire component (Meta Title Assistant, Meta Description Assistant, Content Analysis, Schema Generation, Hreflang Suggester).

If you extended one of these components and redeclared either property with a different type (for example public string $error = ''), Livewire's hydration will now throw a TypeError. Remove the redeclared property from your subclass — the trait already provides it with the correct type.

New Config Keys

Add these blocks to your published config/seo.php if you want to opt into the new surfaces (or leave them at their defaults):

• seo.llms_txt.route_enabled (default false) — serve /llms.txt from the package.
• seo.llms_txt.route_path (default llms.txt).
• seo.feeds — full block (enabled, route_enabled, rss_path, atom_path, title, description, per_page, cache_ttl, feed_id). Bind a FeedProviderContract implementation to the container to supply entries.
• seo.indexnow.route_enabled (default false) — serve the /{key}.txt key-verification file when the filename matches the bound key provider.
• Expanded seo.robots.ai_crawlers (added in this line): the default_allow flag is now a real kill switch (see the fix note above), so if you flip it to false every group without an explicit blocked => false will receive a Disallow: / directive.

Route Wiring

The three new surfaces below register only when their config toggle is true — opt-in so that consumers who serve them via a custom controller are not double-registered:

• GET /llms.txt — LlmsTxtController@index
• GET /feed.xml — FeedController@rss
• GET /feed.atom — FeedController@atom
• GET /{key}.txt — IndexNowKeyController@show

FeedController and IndexNowKeyController both return 404 when no provider (FeedProviderContract, IndexNowKeyProviderContract) is bound in the container.

OG images resolve through the storage disk configured in seo.og_image.disk (default public) via Storage::url() — no route is registered by the package.

No Migration Required

No new database tables, columns, or indexes; no existing model signatures have changed.

New Filter Hooks

• ap.seo.feedEntries($entries, $feedType) — mutate the entry list before RSS/Atom output.
• ap.seo.llmsTxtEntries($entries) — mutate the entry list before the llms.txt manifest is rendered.

Both use the array-of-arrays convention: return whatever the callback accepts, filtered or extended.