VigIA — Full changelog

For each release, see the entries below. The latest version is always at the top.
The current release notes also live in readme.txt under "== Changelog ==".

= 2.5.0 =
* New: A "Database optimization" control in Data settings creates the indexes the statistics tables need. VigIA adds them by itself in the background shortly after updating; the button is there for sites where WP-Cron is disabled or barely runs, and it reports when the work is done. You can close the page while it runs, it finishes on its own.
* Improved: The analytics screens no longer bog down on sites with a long history. The most crawled pages and top crawlers tables were resolving the per-row details for every page in the range instead of the ten on screen, and the pages endpoint re-derived the content type of up to 2000 old rows on every single load. On a test site with 610,000 recorded visits, the whole dashboard went from 16.4 seconds to 1.4 for a year of data, and from 17.1 to 3.6 for the full history.
* Improved: Date ranges longer than 60 days are now computed ahead of time in the background and served ready-made, so the long views open instantly. Shorter ranges are never cached: what you see for today or the last 30 days is always live.
* Improved: Exporting activity to CSV is far lighter. A week of activity used to run more than 15,000 database queries and now runs 12.
* Improved: Visits still waiting for their content type are filled in by a background task that drains them steadily, instead of on whichever page load happened to run into them.
* Fix: Paging backwards through Recent activity left the numbering stuck on the last page loaded from the server, because pages served from the browser's own cache did not update it. The rows changed but the counter did not follow.
* Fix: Rows could show up twice, or not at all, when paging through the activity table, the most crawled pages or the top crawlers. Entries that tied on the sort column had no defined order, so they could change places between one page and the next. The same applies to a CSV export, which is paged the same way.
* Fix: The analytics tables were unreadable on a phone: they were squeezed into the available width until words broke one letter per line. They now keep readable columns and scroll sideways within their box.
* Fix: The spinner on the export button never spun and sat misaligned with the label.
* Fix: CSV exports now write every recorded value as plain text, so spreadsheet software never interprets the contents of a cell.

= 2.4.5 =
* Fix: The llms.txt references in robots.txt could end up duplicated, with a stray "Disallow: /" left in front of the old block, when another plugin rewrote the file and joined two of its lines. Because that leftover sits inside whichever User-agent group precedes it, it could tell every crawler to stay away from the whole site. VigIA now recognises its own markers even when another plugin has glued them to the line above, and repairs an affected robots.txt during the update.
* Fix: robots.txt is now always left with a single newline at the end, so the next plugin that appends to it starts on its own line instead of running into VigIA's last one.
* Fix: The AI crawler rules block is now closed with an end marker. Without it, rules another plugin wrote immediately after the block could be removed along with it.
* Fix: An empty themed icon, such as the ones many themes and page builders emit, no longer leaves a stray ** or * in the middle of a sentence in the Markdown for agents output.
* Fix: Bracketed prose such as [sic] or a [1] footnote is no longer mistaken for a leftover shortcode and deleted from the Markdown output. Tags left behind by a page builder are still removed.
* Fix: The description in the Markdown frontmatter no longer runs words together on tables and lists, now decodes typographic punctuation instead of showing it as entities, and no longer comes out empty on long accented text.

= 2.4.4 =
* Improved: The content type pickers for Markdown for agents and llms.txt now offer only the types these files can actually serve. Types with no address of their own on the front end are gone from the list, and so are the course and membership types whose own plugin decides who may read them, with a note on screen explaining why.
* Improved: A new `vigia_content_is_public` filter lets a plugin or theme decide, per entry, whether it may appear on these files at all.
* Fix: Markdown for agents and llms.txt could serve content that a course or membership plugin keeps behind its gate. Both rebuild an entry's content outside the normal template, where the access rules those plugins enforce in their own templates never run, so an anonymous visitor could read through the .md file what the HTML page withheld. Entries are now checked against the access rules before they are published on any of these surfaces, and the content types whose plugin decides who may read them are withheld unless you opt them in. Your llms.txt and llms-full.txt are rebuilt during the update.

= 2.4.3 =
* Improved: The crawler breakdown of the Most Crawled Pages table now groups its chips by crawler type, each carrying the category colour used elsewhere in the dashboard, so the categories cluster together instead of scattering. Within each category the chips keep their by-visits order.
* Fix: A static front page now advertises its Markdown version. Its permalink is the site home, so there was no path left to suffix with .md and neither the <link rel="alternate"> tag nor the Link header were emitted; both now point to the page slug endpoint, which was already serving that markdown.
* Fix: The page assigned as the Posts page in Settings > Reading no longer serves markdown. WordPress renders the blog loop there instead of the page content, so its .md endpoint returned something the site never displays, and on most installs it was empty.

= 2.4.2 =
* Improved: When Visibility manages the robots.txt rules for AI crawlers, the Disallow & Blocking tab no longer shows the inert Add rule form and Remove buttons: the rules list is read-only while the compliance monitor and the PHP (403) blocking stay fully active.
* Fix: The "Markdown for Agents" pointer in the AI Discovery JSON-LD now declares its URL pattern as an EntryPoint urlTemplate (with text/markdown contentType) instead of a plain target string, so crawlers no longer fetch the literal {slug}.md placeholder and Google Search Console stops reporting it as a 404.

= 2.4.1 =
* Improved: The Visibility recommendation now shows across the three Extras tabs (LLMs.txt, Markdown and JSON-LD) and the AI Visibility analyzer whenever Visibility is not active, with broader SEO plugin detection and clearer copy about the traffic and authority you gain. The call to action is now an inline link instead of an intrusive button.
* Improved: The Monitored AI crawlers list is now grouped by crawler type, so the coloured category badges cluster together instead of scattering across the grid. The analytics tables (top crawlers, recent activity) keep their by-visits order.
* Fix: The analyzer no longer offers to install NoIndexer or Native Sitemap Customizer, which are no longer in the WordPress.org directory (their install modal opened empty).

= 2.4.0 =
* Improved: With Visibility active, VigIA's overlapping Extras tabs (LLMs.txt, Markdown and JSON-LD) are dimmed and their controls disabled, with a pointer to Visibility where you configure them, so the same signal is never edited in two places. They are never removed: a site without Visibility sees them in full. The Disallow & Blocking tab stays active because its compliance monitor and 403 blocking remain VigIA's.
* Improved: The robots.txt rules for AI now read from Visibility when it owns them. VigIA's compliance monitor and 403 escalation check against the crawlers Visibility disallows, and VigIA's own rule editor turns read-only while ceded, so the two never disagree on which bots should be blocked.
* Improved: Each signal tab and the recommendations now point to Visibility, the sibling SEO and AI plugin that pairs natively with VigIA, instead of competing SEO plugins. You get a complementarity note when both are active, and a clear reason to switch (lightweight and native, no bloat) when another SEO plugin is running without Visibility.

= 2.3.0 =
* New: Coexistence with Visibility, the AyudaWP sibling that, from its 1.7.0, generates the AI and search signals (reusing VigIA's code). The family splits by nature, not by topic: Visibility emits, VigIA observes and controls. When Visibility is active and emitting a signal, VigIA automatically steps back from emitting the same one to avoid duplicates, across all four overlap points: Site Identity schema (JSON-LD Organization/Person + WebSite), llms.txt and llms-full.txt, Markdown for agents (/{slug}.md), and the robots.txt rules for AI crawlers. Detection is read from Visibility's own settings, so the handover is per signal: if Visibility has a given signal turned off, VigIA keeps emitting it, so a signal is never dropped by both at once
* New: Each overlapping admin screen (JSON-LD, LLMs.txt, Markdown, Disallow & Blocking) shows a neutral coexistence notice explaining who handles what, instead of the third-party "duplicate schema" warning, which is reserved for SEO plugins that duplicate without coordinating
* New: When VigIA cedes llms.txt, it removes its own physical llms.txt / llms-full.txt so a leftover file at the site root can't shadow Visibility's virtual output. It recognises Visibility's own files by their "# Generated by Visibility" header and never deletes them, and only removes files VigIA itself generated. When it cedes the robots.txt rules for AI, it withdraws its block from the physical robots.txt (marker-scoped), leaving the rest of the file untouched
* New: Developer filter vigia_defer_emission_to_visibility (default true, passed the signal name) to override the handover and keep VigIA's own emission even when Visibility is active and emitting that signal
* Improved: VigIA's core is untouched by the handover. Crawler analytics, stats, hard PHP/403 blocking, email alerts and the control MCP server keep working exactly as before. Its robots.txt block-by-crawler editor yields to Visibility, but the real enforcement (PHP/403) stays with VigIA, which is its differentiator

= 2.2.0 =
* New: The "Most crawled pages" table is now an analysis hub for each page. It adds a Content type column (the dominant type per page, resolved with a single query reusing the value detected and stored since 2.0.0), a Trend arrow comparing crawler visits against the previous period (with a "Before → Now" tooltip; shown as "new" for pages with no prior traffic and hidden for the all-time range), and an expandable Crawlers column that loads, on demand, the per-bot breakdown for that exact URL
* Improved: The "Top crawlers" table now shows each bot's last visit ("2 hours ago") and a pages/visits ratio (unique pages over total visits), so you can tell whether a crawler is still active and whether it explores your site broadly or keeps hitting the same handful of URLs
* Improved: Expanded and refreshed the AI crawler catalog to 60+ known crawlers, each checked against the operators' own documentation. Added OAI-AdsBot (OpenAI), Claude-Code (Anthropic), Google-NotebookLM and Google-Read-Aloud (Google), Amzn-SearchBot, Amzn-User and bedrockbot (Amazon), MistralAI-Index (Mistral), SBIntuitionsBot (SB Intuitions), Kimi-User (Moonshot AI) and YandexAdditional (Yandex). Corrected several categories and operators: FacebookBot and TikTokSpider are AI training, Gemini-Deep-Research and Cohere-AI are assistants, Google-CloudVertexBot and PanguBot are data scrapers, Meta-WebIndexer and Cloudflare-AutoRAG are AI search, PanguBot belongs to Huawei, ICC-Crawler to NICT, and Omgili and Webzio-Extended to Webz.io
* Improved: Refreshed the bundled MCP server library (WordPress MCP Adapter) to its latest upstream. The notable fix is in the tool validator: tools that take no parameters (whose schema properties come through as an empty object) now validate correctly. VigIA ships a couple of parameter-less abilities, so this hardens its MCP server
* Fix: Removed four crawler tokens that no operator actually uses (Groq-Bot, Together-Bot, Replicate-Bot, HuggingFace-Bot) and corrected xAI's crawler token from the non-existent "xAI-Bot" to the real "GrokBot"
* Fix: Hardened the admin dashboard against stored XSS. The escapeHtml() helper was using the textContent/innerHTML DOM pattern, which only neutralises &, <, > — not quotes. A request_path containing a literal " (delivered via a raw socket or curl --path-as-is with a known crawler User-Agent) would survive sanitize_text_field(), persist in the database, and break out of quoted HTML attributes (data-path, title, href) when an admin opened the dashboard. Rewritten to use explicit .replace() chains for &, <, >, " and '

= 2.1.0 =
* New: Command Palette actions. The WordPress command palette (Cmd/Ctrl+K, admin-wide since WP 6.9) gains two VigIA actions — "Regenerate llms.txt now" and "Regenerate llms-full.txt now" — that rebuild each file from your saved configuration without leaving the current screen. Navigation to VigIA's screens is intentionally left to WordPress, which already lists every admin menu page in the palette, so the plugin only adds the actions WordPress can't generate on its own
* Improved: The VigIA admin menu now sits at the bottom of the admin menu (position 999), separated from the content menus, instead of wedging between the content menus near the top
* Improved: Refreshed the bundled MCP server libraries — the WordPress MCP Adapter and php-mcp-schema — to their latest upstream versions, so the self-contained MCP server ships current upstream fixes
* Fix: Markdown for Agents was emitting links as a bare "(url)" with no anchor text, and images as "!(url)" with no alt text. Links and images were converted correctly, but the final "shortcode artifact" cleanup pass matched the bracketed label of every `[text](url)` and `![alt](url)` and deleted it. The cleanup now uses a negative lookahead so it only strips genuine leftover shortcodes, and inline code and code blocks are protected so brackets such as `$arr[key]` survive
* Fix: Markdown for Agents list and code handling rewritten. Nested lists now keep each sub-item's bullet/number and are indented per level (walked with DOMDocument instead of the old flat regexes that dropped the first nested item's marker and all indentation), bold and italics are preserved inside numbered lists (they used to be stripped, unlike bulleted lists), pipes inside table cells are escaped so a cell can't break its column, and the language of a fenced code block is detected from a `language-xxx` class on either the `<pre>` or the `<code>` element
* Fix: Markdown for Agents inline formatting no longer leaks stray `**` / `*` markers. The bold/italic/strikethrough conversion matched the start of any tag beginning with the same letter — `<b>` also caught `<button>`/`<br>`, `<i>` caught `<img>`/`<iframe>`, `<s>` caught `<span>`/`<svg>`/`<section>` — so the `<button>` WordPress injects for the image lightbox paired with a later `</strong>` and scattered markers through the output. Tag matching now requires a word boundary after the tag name
* Fix: Internal security hardening to align with the WordPress.org plugin review guidelines. The JSON-LD is now emitted through `wp_print_inline_script_tag()` (with `<` and `>` escaped in the encoded JSON so data can't break out of the inline script), and the Extras page validates the requested tab against an allowlist

= 2.0.3 =
* Improved: Internal security hardening to align with the WordPress.org plugin review guidelines.
* Fix: Prevented a fatal error ("Call to a member function wp_rewrite_rules() on null") that could occur when an AI crawler hit was tracked on `shutdown` during an early-exit request, such as a 301 redirect handled before WordPress initializes its rewrite rules.

= 2.0.2 =
* Fix: The recent activity table now records the real HTTP status of each AI crawler hit. Until now every visit was stored as HTTP 200, even when the crawler actually received a 404, 301 or 410, because the status was read on the `init` hook — early in the request, before WordPress resolves the query and sends the 404/redirect status. The visit is now logged on the `shutdown` hook, where `http_response_code()` reflects the status WordPress actually sent. The same root cause was mistyping not-found hits in the Content type column, since the classifier keys "404 Not found" off the stored status; a 404 hit is now typed correctly. Note: rows recorded before this update keep their stored 200 — only visits logged from 2.0.2 onward carry the corrected status
* Fix: A crawler request served by a Markdown for Agents `.md` endpoint is no longer logged twice. The endpoint records its own hit and then exits the request; the general tracker, now running on `shutdown`, would otherwise log the same request again. The tracker now skips any request that has already been recorded

= 2.0.1 =
* Fix: Markdown for Agents and llms.txt no longer expose password-protected content. Previously, when the feature was enabled, a published-but-password-protected post or page was served in full through its `.md` URL (and via `Accept: text/markdown` content negotiation), bypassing the WordPress password form; the same content also leaked as title + excerpt inside taxonomy term `.md` listings. The `.md` endpoints now return 404 for any post or page with a password set (`is_post_eligible` and `find_post_by_path` honor `post_password`, and the slug lookup uses `has_password => false`), password-protected posts are dropped from the per-term post listing (`get_term_posts_section` uses `has_password => false`), and those entries are excluded from llms.txt / llms-full.txt listings and counts
* Fix: Hardened deserialization of plugin options and post meta against PHP object injection. The three `maybe_unserialize()` sinks (llms.txt settings, robots.txt llms reference, SEO noindex meta) now decode with `unserialize( $data, array( 'allowed_classes' => false ) )` after an `is_serialized()` check, so no object can be instantiated from stored data

= 2.0.0 =
* NEW: Markdown for Agents now serves taxonomy archive pages. Categories, tags, WooCommerce product categories and any custom public taxonomy can be exposed as `.md` endpoints with their own YAML frontmatter (title, description, url, type: term, taxonomy, parent, count, image, lang), term description rendered through the_content filter, list of direct child terms for hierarchical taxonomies and an excerpt of the latest posts/products assigned to the term. Disabled by default — opt in per taxonomy in VigIA > Extras > Markdown for Agents
* NEW: WooCommerce product `.md` endpoints now include schema-like fields in the YAML frontmatter (sku, product_type, price, regular_price, sale_price, currency, availability, stock_quantity, rating, rating_count, review_count). The product listing inside a `product_cat` term archive also gets an inline summary with formatted price, "was X" when on sale, star rating and an out-of-stock flag
* NEW: Term-specific URL like `/category/electronics.md` and `Accept: text/markdown` content negotiation on taxonomy archive pages. `<link rel="alternate" type="text/markdown">` HTML tag and Link HTTP header are added on enabled term archives so AI agents discover the markdown version
* NEW: Advanced filters on the recent activity table — multi-select crawler picker, content type filter (post, page, product, category archive, tag archive, date/author archive, feed, sitemap, REST API, file, other), HTTP status filter (200, 301, 304, 403, 404, 410, other), and a configurable date range (from/to inputs). The "Export filtered CSV" button downloads exactly what the filters return and an active-filter counter signals when any filter is in effect
* NEW: Two new visible columns on the recent activity table — Content type and HTTP status. HTTP cells are color-coded (2xx green, 3xx blue, 4xx/5xx red) for quick scanning
* NEW: Content type detection now distinguishes four extra buckets that used to fall under "Other": "Home" (the / path), "404 Not found" (any hit with http_status 404, regardless of path), "Admin / login attempt" (/wp-admin and /wp-login.php — useful to spot bots probing the admin) and "WordPress system" (admin-ajax.php, xmlrpc.php, wp-cron.php, wp-comments-post.php). DB version bumped to 1.2.0; the backfill cron revisits rows previously marked "other" so they end up in the right bucket without manual intervention
* NEW: Server-side pagination on the recent activity table. The page size stays at 20 but the count, navigation and CSV export now operate over the full database instead of the last 500 rows in memory — works the same on sites with millions of hits. Four-button pager (first, previous, next, last) mirrors the rest of the dashboard tables. A small client-side cache keeps the last 3 pages warm so revisiting them does not hit the server again
* NEW: Public `VigIA_Database::query_visits( $args )` for filtered/paginated queries and `VigIA_Database::detect_content_type( $path )` plus `get_content_type_options()` helpers. The `/vigia/v1/recent` and `/vigia/v1/export` endpoints accept new query params: `crawlers[]`, `category`, `content_type`, `http_status`, `date_from`, `date_to`, `page`, `per_page`. The legacy contract — no parameters returns the latest 500 visits as a flat array — is preserved
* NEW: Filter `vigia_markdown_term_eligible` (mirror of `vigia_markdown_post_eligible`) to programmatically allow or deny term markdown output. Filter `vigia_markdown_term_posts_limit` to tune the per-term post listing size (default 20)
* NEW: Every CSV export (activity, filtered activity, timeline summary) now opens with a metadata banner — site name, site URL, export type, date range, export timestamp, generator string and (when filters are active) a summary of the applied filters. The activity export uses the `vigia-filtered-YYYY-MM-DD.csv` filename when any filter is in effect, and `vigia-YYYY-MM-DD.csv` otherwise
* DB: New `content_type` column on `wp_vigia_visits` (DB version bumped to 1.1.0). Populated at insert time via `detect_content_type( $path, $http_status )` which combines structural patterns (feed/sitemap/REST/file/category/tag/archive/admin/wp-system) with a `url_to_postid()` lookup so post / page / product / CPT visits land in the right bucket. Pre-2.0.0 rows are backfilled hourly by the `vigia_backfill_content_type` cron in batches of 500 until the queue is empty; when the user filters by content_type the next 2000 unprocessed rows in the visible range are reclassified synchronously
* Changed: Admin menu order restored. Analytics is the default landing page again when clicking the VigIA top-level menu, with AI Visibility in second position. The plugin row description on the Plugins screen is unchanged
* Changed: Per-term noindex settings from Yoast SEO, Rank Math, All in One SEO and SEOPress are honored when "Respect LLMs.txt exclusion filters" is enabled. Single-post noindex detection also recognises Native SEO NoIndexer (AyudaWP) via `Noindexer_Frontend::is_noindex()`
* Changed: Activity-table filter row no longer includes the URL and IP text inputs (they were not part of the upgraded server-side workflow). Filters now applied: Crawlers, Category, Content type, HTTP status, Date range
* i18n: All new strings (filter labels, content-type options, badge texts, multiselect placeholders, server-side pager range, CSV banner labels) ship translation-ready. CSV export headers (Crawler, Category, Page, IP Address, HTTP Status, Content type, Date) all wrapped in `__()`
* Dev: Markdown response writer refactored into a shared `send_markdown_response()` private method so post and term responses share the blocker check, analytics tracking and header logic

= 1.12.1 =
* Fix: Fatal error "Cannot redeclare WP\MCP\constants()" when both VigIA and the standalone `mcp-adapter` plugin are active. VigIA now skips its bundled adapter bootstrap if the standalone copy has already been loaded

= 1.12.0 =
* NEW: One-click MCP setup. The MCP Adapter and php-mcp-schema dependencies now ship bundled inside the plugin, so the server is active right after install — no `composer install` and no terminal required
* NEW: "Quick connect" panel in VigIA > Extras > MCP that generates a dedicated `VigIA MCP` Application Password and renders ready-to-paste connection commands for Claude Code, Cursor, Claude Desktop and a generic block (URL + Authorization header) that fits any other MCP client (Codex CLI, Continue, Cline, Antigravity, Zed, etc.)
* NEW: Safe JSON merger for Cursor and Claude Desktop. Paste your current config file, the plugin parses it, splices VigIA into `mcpServers` preserving everything else (preferences, other servers) and returns a valid file ready to save back, with no manual comma juggling
* NEW: Detection and revocation flow for an existing `VigIA MCP` Application Password directly from the MCP tab
* NEW: "What can I ask my AI now?" section in the MCP tab with example prompts so users immediately see what the connection unlocks
* NEW: Read-only mode is now a one-click toggle in the MCP tab (option `vigia_mcp_read_only`). The `vigia_can_write_via_abilities` filter still works for developers who prefer to force it from code and takes precedence over the toggle
* Changed: Claude Desktop snippet now launches `mcp-remote` via `npx` as a stdio-to-HTTP bridge (the only way Claude Desktop can currently talk to a remote HTTP MCP server). Requires Node.js installed on the machine. Claude Code and Cursor speak HTTP MCP natively and need no bridge
* Changed: Manual setup snippets moved into a collapsed "Manual setup (advanced)" block to avoid competing with Quick connect; inactive-server message now references a missing `vendor/` directory instead of a missing Composer install
* Fix: `vigia/get-blocked-items`, `vigia/get-top-pages` and `vigia/block-crawler` abilities were calling non-existent methods and returned a 500 when invoked. They now resolve correctly and the output surfaces fields that were being discarded (`name` for blocked items, `crawler_count` for top pages)

= 1.11.0 =
* NEW: Native MCP server that exposes VigIA abilities at /wp-json/vigia/v1/mcp through the official WordPress MCP Adapter, ready to plug into Claude Code, Cursor or Claude Desktop
* NEW: MCP tab in VigIA > Extras with adapter status, endpoint URL, Application Password setup, connection snippets for Claude Code, Cursor and Claude Desktop, list of exposed abilities and the read-only mode snippet
* NEW: Filter `vigia_can_write_via_abilities` to disable mutating abilities (block, unblock, robots changes) while keeping read-only abilities working - useful when granting MCP access with Application Passwords
* Fix: `vigia/block-crawler` now validates IP addresses with FILTER_VALIDATE_IP, matching the behavior of the AJAX blocking handler
* Performance: results of `vigia/get-crawler-stats`, `vigia/get-top-crawlers` and `vigia/get-top-pages` are now cached for 5 minutes via transients to keep MCP clients from hammering the database
* Changed: Version history moved to a dedicated changelog.txt file served from the plugin's public SVN, keeping readme.txt focused on the current release
* Dev: introduced composer.json declaring the optional `wordpress/mcp-adapter` dependency

= 1.10.2 =
* Fix: Removed stray whitespace before PHP opening tag in class-visibility-analyzer.php that caused "headers already sent" warnings and could break other plugins functionality

= 1.10.1 =
* Fixed: AI Visibility Analyzer now detects sitemaps.xml (plural) in addition to sitemap.xml, sitemap_index.xml and wp-sitemap.xml

= 1.10.0 =
* Added: NoIndexer plugin integration for llms.txt generator, Markdown endpoints, and AI Visibility Analyzer
* Added: Posts marked as noindex by NoIndexer are now automatically excluded from llms.txt and .md endpoints
* Added: NoIndexer detection displayed in LLMs.txt Generator settings alongside SEO plugin detection
* Added: NoIndexer detection and recommendation in AI Visibility Analyzer
* Added: NoIndexer to promotional banner catalog
* Updated: Periscope plugin renamed from Periscopio in promotional banner catalog

= 1.9.5 =
* Tested up to WordPress 7.0
* Updated: Promotional banner catalog with latest plugins and services

= 1.9.4 =
* Added: Google-Agent, PetalBot, and Applebot crawlers to built-in detection list
* Improved: Table headers now display "URL" instead of "Page" for clarity
* Improved: Better contextual links in Extras page (FAQ reference for blocking, updated Markdown for Agents documentation)

= 1.9.3 =
* Improved: AI Visibility Analyzer now distinguishes between perfect (Excellent, green) and good (Good, blue) scores for clearer visual feedback at a glance (Props to @fpuenteonline)

= 1.9.2 =
* Fix - Robots.txt: fixed VigIA rules merging with existing content from other plugins (e.g., Sitemap lines) due to missing line breaks

= 1.9.1 =
* Improved: Bottom pagination controls added to recent activity table for easier navigation on long lists
* Fixed: Blocking a crawler from recent activity no longer resets pagination to page 1

= 1.9.0 =
* NEW: Client-side paginated navigation replaces "Load more" buttons across all tables
* NEW: Reusable VigiaPaginator component with first/prev/next/last arrows and range indicator (e.g., 1–10 of 85)
* NEW: Analytics page tables: Top crawlers (10/page), Most crawled pages (10/page), Recent activity (20/page), Custom crawlers (5/page)
* NEW: Extras page tables: Disallow rules, Compliance check, Blocked User-Agents, Blocked IPs (5/page each)
* NEW: Arrows only shown when multiple pages exist, range always visible
* Improved: Remove/Unblock actions in Extras page now update inline without page reload (smooth fade animation)
* Improved: Add Disallow rule in Extras page now inserts row and updates robots.txt preview without reload
* Improved: Block via PHP from compliance panel updates button state without reload
* Improved: Pagination automatically recalculates when rows are added or removed
* Improved: Top crawlers and top pages now load all results in a single request (client-side pagination)

= 1.8.1 =
* NEW: Clickable URLs in the Recent Activity table (opens in new tab, matching Top Pages behavior)
* NEW: Neutral "info" status icon for checks that are not applicable (e.g., no images on page, share buttons on homepage)
* NEW: Periscopio plugin added to the promotional banner catalog
* Improved: Visibility Score banner redesigned with outlined style - subtle tinted backgrounds and colored text instead of solid color blocks
* Improved: WordPress 7.0 compatibility for all button styles (dashicons alignment in Generate, Export CSV, Clear, Home, and Re-analyze buttons)
* Improved: WordPress 7.0 compatibility for promotional banner buttons (min-height override)
* Improved: AI Share & Summarize recommendation now appears on homepage analysis when plugin is not installed or active
* Improved: Image alt text check shows neutral "info" status when no images are found instead of a misleading green pass
* Improved: AI share buttons check shows neutral "info" status on homepage instead of a misleading green pass
* Fixed: Comparison chart not updating when changing period with comparison enabled
* Fixed: Image alt text parser incorrectly matched `data-wp-bind--alt` (WordPress Interactivity API) as real alt attributes
* Fixed: Image alt text parser now correctly counts `alt=""` (empty alt for decorative images) as valid
* Fixed: Removed dead CSS rule for export button icon (vertical-align inside flex container)

= 1.8.0 =
* NEW: AI Visibility Analyzer - 100-point scoring system with letter grades (A+ to F)
* NEW: 20 individual checks across 5 weighted categories
* NEW: Access & AI Discovery checks (37 pts): robots.txt, AI bot directives, Content Signals, llms.txt, llms-full.txt, sitemap, RSS feed, noai meta
* NEW: Structured Data & Semantic Context checks (25 pts): JSON-LD types, Open Graph, Twitter Cards, meta description, canonical URL, language
* NEW: Content Structure & Readability checks (20 pts): H1, heading hierarchy, semantic HTML5, image alt text, content/HTML ratio, JS independence
* NEW: AI Interaction & Distribution checks (8 pts): markdown delivery, AI share buttons
* NEW: Access Performance checks (10 pts): TTFB measurement with tiered scoring
* NEW: Smart recommendations with 4 tiers: VigIA features, AyudaWP plugins, inactive plugins, third-party SEO
* NEW: URL autocomplete selector to analyze any page on your site (posts, pages, custom post types)
* NEW: Results cached for 24 hours with manual re-analyze option
* NEW: AI Score link in plugin action links (plugins page)
* NEW: Dedicated VigIA > AI Score admin page

= 1.7.0 =
* NEW: JSON-LD structured data generator with Site Identity and AI Discovery sections
* NEW: Site Identity - WebSite and Organization/Person schema with sameAs social profiles
* NEW: AI Discovery - ReadAction pointers to llms.txt, llms-full.txt, and Markdown for Agents endpoints
* NEW: SearchAction support for Google sitelinks search box
* NEW: Media library integration for JSON-LD logo selection
* NEW: JSON-LD conflict detection warns when SEO plugins already generate WebSite/Organization schema
* NEW: JSON-LD output page selector (front page or any published page)
* NEW: Live JSON-LD preview with real-time updates
* NEW: JSON-LD automatically detects available LLMs.txt and Markdown for Agents features
* NEW: JSON-LD tab in Extras page

= 1.6.1 =
* Improved: New plugin suggestion added.

= 1.6.0 =
* NEW: AI Share & Summarize integration - see share button clicks per page in the Most Crawled Pages table
* NEW: Clickable URLs in the Most Crawled Pages table (opens in new tab)
* NEW: Cross-plugin promotion tip for AI Share & Summarize (dismissible)
* Improved: Most Crawled Pages table dynamically shows Clicks column when AI Share & Summarize is active
* Fixed: Action button icons and dropdown arrow now properly centered vertically

= 1.5.0 =
* NEW: Markdown for Agents - serve individual posts as optimized markdown for AI agents
* NEW: Dedicated .md URL endpoints for each post and page
* NEW: Accept: text/markdown content negotiation support
* NEW: Discoverability via link rel="alternate" HTML tags and Link HTTP headers
* NEW: YAML frontmatter with full post metadata (title, date, author, categories, tags, image, language)
* NEW: X-Markdown-Tokens response header
* NEW: Respects LLMs.txt exclusion filters (noindex, URL patterns, manual excludes)
* NEW: Blocked crawlers receive 403 on markdown endpoints too
* NEW: Analytics integration tracks markdown requests from known crawlers
* NEW: Configurable post type selection for markdown endpoints
* NEW: New "Markdown for Agents" tab in Extras page
* Follows the Cloudflare Markdown for Agents standard

= 1.4.1 =
* Fixed: Abilities API category registration error on WordPress 6.9+
* Improved: Abilities API initialization method for better compatibility
* Improved: Dynamic promotional banners with random plugin and service rotation

= 1.4.0 =
* NEW: WordPress Abilities API integration (requires WordPress 6.9+)
* NEW: 9 abilities for AI agents and automation tools
* NEW: Analytics abilities: get-crawler-stats, get-top-crawlers, get-top-pages
* NEW: Blocking abilities: get-blocked-items, block-crawler, unblock-crawler
* NEW: Robots.txt abilities: get-robots-rules, add-robots-disallow, remove-robots-rule
* NEW: Custom ability category "AI Crawler Analytics" for better discoverability
* Improved: Graceful degradation for WordPress versions below 6.9

= 1.3.0 =
* NEW: Period indicator in dynamic section titles (timeline, categories, crawlers, pages)
* NEW: "Load more" pagination in Top crawlers and Top pages tables (10 items per load, up to 100)
* NEW: Showing counter display (e.g. "Showing 10 of 47")
* Improved: Plugin recommendation boxes now available on Extras page
* Fixed: Shortcodes from page builders and common plugins now display as clean text in llms.txt files

= 1.2.9 =
* Fixed: robots.txt Disallow rules now work correctly with physical robots.txt files (Yoast SEO and other plugins that create physical files)
* Improved: AI crawler rules are now synced to physical robots.txt automatically when adding or removing rules

= 1.2.8 =
* Fixed: UTF-8 encoding issues in llms.txt and llms-full.txt files (special characters like accents now display correctly in all browsers)

= 1.2.7 =
* Fixed: "Learn more" link is now translatable

= 1.2.6 =
* Fixed: LLMs.txt Generator now correctly adds robots.txt references when saving and generating
* Improved: Simplified LLMs.txt Generator to single "Save and generate" button for clearer workflow
* Updated: "Learn more" link now points to AyudaWP documentation

= 1.2.5 =
* Fixed: Responsive layout issues with charts and content boxes overflowing on smaller screens
* Improved: Better breakpoint handling for stats grid and charts adaptation

= 1.2.4 =
* Fixed: Renamed duplicate "VigIA" submenu to "Analytics" for better navigation clarity

= 1.2.3 =
* Fixed: Activation notice now properly dismisses when closed on the plugins page

= 1.2.2 =
* Improved: Added PHPCS ignore comments for intentional direct database queries (cache bypass)
* Improved: Added translator comments for all placeholder strings in LLMs generator

= 1.2.1 =
* Fix: LLMs.txt generator now correctly handles taxonomy filters for each post type
* Fix: Taxonomy filters (categories, tags) no longer incorrectly exclude pages or other post types
* Fix: Resolved object cache conflicts on hosts with Memcached/Redis (SiteGround, etc.)
* Fix: Settings now persist correctly after saving and regenerating
* Fix: Robots.txt references only added when llms files actually exist
* Fix: Robots.txt integration reads fresh data bypassing object cache
* Improved: Added "Save settings" button to save options without generating files
* Improved: Direct database queries for settings to avoid cache inconsistencies
* Improved: Better boolean normalization for checkbox values

= 1.2.0 =
* NEW: Redesigned LLMs.txt generator with improved content selection
* NEW: Select content by post type (posts, pages, custom post types)
* NEW: Filter content by taxonomies (categories, tags, custom taxonomies)
* NEW: Manual include with AJAX-powered search
* NEW: Manual exclude with AJAX-powered search
* NEW: Exclude content by URL patterns with wildcard support
* NEW: SEO plugin integration - auto-exclude noindex content
* NEW: Supported SEO plugins: Yoast, Rank Math, AIOSEO, SEOPress, The SEO Framework
* NEW: Auto-regeneration options (daily, weekly, monthly)
* NEW: Add llms.txt and llms-full.txt references to robots.txt
* Improved: Scalable architecture for sites with thousands of posts
* Improved: Better UI/UX for content selection

= 1.1.1 =
* Fix: Version dump to improve texts.

= 1.1.0 =
* NEW: PHP-based crawler blocking (403 Forbidden response)
* NEW: Robots.txt management for AI crawlers
* NEW: Compliance monitoring (detect crawlers ignoring robots.txt)
* NEW: Email alerts system (daily/weekly/monthly reports)
* NEW: LLMs.txt and llms-full.txt file generator
* NEW: Extras page with all new management tools
* NEW: Quick block dropdown in crawlers table
* Improved: Block actions directly from analytics dashboard

= 1.0.0 =
* Initial release
* AI crawler detection and tracking
* Analytics dashboard with charts
* CSV data export
* Custom crawler support
* Dashboard widget