`) instead of giving up. Sites where `post_content` already holds real text are unaffected — this fallback path never runs for them. #### 0.7.5 * Fixed: llms.txt’s plain-text excerpt fallback (used when no AI-enhanced description is cached) could reduce a body of many empty paragraphs (` ` entities) into a lone ellipsis (“…”) instead of no description at all — `wp_trim_words()` was counting each undecoded ` ` as a separate word and appending its truncation marker. Entities are now decoded before word-counting. #### 0.7.4 * Fixed: AI-enhanced schema generation still treated a post/page body consisting only of empty paragraphs (` ` HTML entities, as some page builders insert) as non-empty content, since `wp_strip_all_tags()` only strips HTML tags, not entities. It’s now decoded and normalized before the empty check, matching llms. txt’s existing behavior. #### 0.7.3 * Fixed: stale AI-enhanced schema descriptions generated by the 0.7.2-and-earlier shortcode-only-content bug are now cleared automatically once the content is correctly recognized as empty, instead of lingering forever. * Fixed: the “AI強化スキーマの更新” (bulk AI schema update) button on the post/ page list screen could hit a 504 Gateway Time-out on sites with many posts, since it processed everything in one synchronous request. It now runs in small AJAX- driven steps until done, matching the FAQ bulk-generate flow. * Added: a new `llmo_sitemap_exclude_post_types` filter for sitemap.xml, for excluding theme/page-builder-registered post types by name when they’re registered as publicly viewable (so `is_post_type_viewable()` alone can’t detect them as non-content). #### 0.7.2 * Fixed: sitemap.xml could include theme/page-builder internal post types (e.g. header/footer/slider parts) that aren’t actually viewable single pages. Now filtered with the same `is_post_type_viewable()` check WordPress core sitemaps use. * Fixed: AI-enhanced schema generation (and llms.txt indirectly, via the cached description) treated a post/page whose body was only a shortcode (e.g. `[geo_plugin_faq]`) as non-empty, sending the raw shortcode text to the AI instead of skipping it. Now shortcodes are stripped before the empty-content check, matching llms.txt’s existing behavior. * Added: llms.txt now falls back to the front page’s AI-enhanced description for its summary line when the site tagline (Settings > General) is empty. #### 0.7.1 * Fixed: FAQ bulk-generate (AI analysis mode) could fail entirely with a validation error when the candidate posts/pages included one with no extractable excerpt text (e.g. image- or embed-only content). Such posts are now excluded from the batch, matching the existing empty-title exclusion. #### 0.7.0 * Added: English (en_US) translation. The plugin’s admin UI now automatically switches to English when the WordPress site language is set to English (Settings > General > Site Language), following WordPress’s standard translation mechanism. No settings UI change — this follows the site-wide language, matching WordPress’s normal behavior for all admin screens. #### 0.6.26 * Changed: shortened the FAQ bulk-generate description text on the FAQ list screen( removed the sentence about specifying a count, kept the rest). #### 0.6.25 * Changed: shortened the sampling investigation description added in 0.6.24 — removed the sentence about execution frequency being fixed by plan (kept only the settings- screen pointer). #### 0.6.24 * Added: a brief explanation under “AI引用サンプリング調査” on the access analysis page, clarifying that execution timing/frequency is fixed by plan (no manual trigger) and that investigation keywords are added/edited from the Settings screen. #### 0.6.23 * Changed: the description text for “AIによるキーワード候補” now stays visible outside the collapsible accordion (previously it collapsed along with the suggestion list). Moved the “追加” (add) button beside the keyword input in “調査項目を追加”. Added bottom margin to the “AI引用サンプリング調査” block to visually separate it from “Google Analyticsとの連携”. #### 0.6.22 * Changed: the “AIによるキーワード候補” (AI keyword suggestions) list in the sampling investigation settings is now a collapsible accordion, expanded the first time you view a given analysis result and collapsed on subsequent visits. Moved “調査項目を追加” (add investigation item) above the suggestions list. #### 0.6.21 * Improved: GEO score’s “サイトを分析する” (analyze site) now sends a wider, excerpt- only page set (up to 40 pages, matching FAQ bulk generation’s site-profiling breadth) for the shared business-understanding step, separate from the 10-page full-content set used for actual page-quality scoring. Previously the business/ category/content-gap analysis was limited to the same 10 pages used for quality scoring, giving it a shallower view of the site than FAQ generation’s equivalent step. #### 0.6.20 * Removed: a dead code path for FAQ site profiling (`profile_site()`‘s call into the `Faq\FaqSiteProfiler` class) that had become unreachable in either generation mode since a 2026-08-13 refactor inverted its calling condition. No user-facing behavior changes — the active site-profiling path (used by both generation modes) is unaffected. #### 0.6.19 * Fixed: AI-suggested keyword candidates (sampling investigation section) were pre-filled with a page-title-like phrase (e.g. “運営者情報（会社概要）”), which reads unnaturally as a search query. The backend now generates a separate `keyword` field per suggestion — phrased as a natural search query/question a real user would type into an AI search engine — distinct from the `topic` field still used for content generation. Falls back to `topic` if the AI omits it. #### 0.6.18 * Fixed: the GEO score “サイトを分析する” (analyze site) action now sends the site’s title and tagline (`get_bloginfo('name')`/`get_bloginfo('description')`) to the backend, matching what FAQ bulk generation’s profiling step already sends. Previously this signal was missing, so on sites with little real content the AI had almost no business context to reason from and tended to fall back on generic suggestions. #### 0.6.17 * Added: AI-suggested keyword candidates in the sampling investigation settings section. When GEOスコア「サイトを分析する」or FAQ bulk generation has already produced a business/content analysis, the AI’s inferred “not yet covered” topic candidates are shown as one-click add-to-investigation suggestions (pre-filled keyword field, both AI services checked by default). Suggestions already present as an investigation keyword are hidden, and the block is hidden entirely once the item limit is reached. #### 0.6.16 * Added: a “GEOスコア（ページ品質評価）” (GEO score / page quality evaluation) section on the access analysis page. Click “サイトを分析する” to have AI evaluate up to 10 of your pages against E-E-A-T and PREP-structure criteria (a proxy for how citable they are to AI search engines). Results are shown as a ranked list; click a row to see strengths, weaknesses, and concrete improvement suggestions for that page. #### 0.6.15 * Added: an “AIクローラーのアクセスログ” (AI crawler access log) section on the access analysis page. Detects and lists real visits from known AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Bytespider, CCBot, Amazonbot) based on their User-Agent, showing hit counts and recent activity per bot. Works regardless of license status— no backend communication or AI API calls are involved. Logs older than 90 days are pruned automatically. #### 0.6.14 * Added: a self-serve sign-up flow (standard plan) on the settings page. If you don’t have a license key yet, you can now sign up directly via Stripe Checkout for “AI引用サンプリング調査” (AI citation sampling) and/or “コンテンツ生成” ( content generation) without going through a reseller. Once subscribed, you can add the other plan or manage/cancel your subscription via a Stripe Customer Portal link, right from the settings page. #### 0.6.13 * Fixed: after approving the Google Analytics consent screen, the browser could fail to reconnect to the site’s own admin page (particularly on local environments using a non-standard port) instead of returning to the settings page with a success notice. #### 0.6.12 * Added: a “Google Analyticsとの連携” (Connect Google Analytics) section on the settings page. Connect your Google Analytics (GA4) account, enter your GA4 property ID, and see your site’s session count for the last 7 days. Read-only access only— this plugin never modifies your Google Analytics settings or data. #### 0.6.11 * Fixed: on the post/page list screen, the “AI強化スキーマの更新” button disappeared entirely whenever every post already had AI-enhanced schema generated (the common case now that generation happens automatically on publish). The button and its description are now always shown, matching the intent of keeping it visible for its promotional value. #### 0.6.10 * Changed: the “AI強化スキーマを一括生成” button on the post/page list screen is now labeled “AI強化スキーマの更新”, reflecting that AI-enhanced schema is already generated automatically whenever a post or page is published or updated — this button is for catching up any remaining ones (e.g. content published before AI mode was enabled). * Changed: the update button is now disabled (and shows “更新中…”) immediately after being clicked, so it is clear that generation is in progress instead of appearing unresponsive. #### 0.6.9 * Changed: on the FAQ bulk generation screen, the “FAQ化に有用な候補が見つかりませんでした。” status message and its “この結果について” explanation now appear directly below the analyze/cancel buttons, instead of after the (often more prominent) site summary accordion. * Fixed (backend): under the newly-parallelized queue workers, a FAQ analysis or generation batch could finish processing every item yet never flip to “completed”, leaving the screen stuck showing “解析中: ○/○” with only “中断” (cancel) able to unblock it. Progress tracking is now race-free. * Fixed (backend): if a previously-generated FAQ draft had been deleted in WordPress and the same post became a candidate again, generating from it could silently finish with “取り込み: 0件のページ（スキップ: 0件） 0件のFAQ投稿” instead of actually creating a new draft. It now regenerates properly. #### 0.6.8 * Changed: on the FAQ bulk generation screen, the site summary and expected FAQ categories are now shown collapsed by default inside an accordion, instead of always visible. * Removed: the raw “解析結果の一覧を表示” candidate list accordion. Its place is now taken by a short status message (e.g. “FAQ化に有用な候補が見つかりませんでした。”) plus a collapsed accordion explaining why, when there is nothing usable to generate from. * Changed: after FAQ generation completes, the screen returns to the pre-analysis state so a fresh analysis can be started again right away. * Changed: when an analysis finds no usable candidates, the “サイトを解析する” button is disabled until you leave and revisit the page (or new content is added and the page is reloaded), instead of staying clickable with nothing useful to do. Revisiting the page always re-enables it. #### 0.6.7 * Fixed: re-running FAQ site analysis after a previous analysis had already completed could keep showing the same stale (and possibly exhausted) candidate list forever, making it look like re-analysis and re-generation were broken. Starting a new analysis now always clears the locally cached candidate list and site-summary cache first. * Fixed: a single post or page with an empty title among the FAQ analysis candidates caused the entire analysis request to fail validation, blocking FAQ analysis for the whole site. Untitled posts/pages are now excluded from analysis instead. #### 0.6.6 * Fixed: the top-right “コピー” copy button next to the open/close toggle now sits flush against the right edge of the button row (previously it sat immediately after the toggle button instead). #### 0.6.5 * Changed: the “生成結果を開く” (open result) button on an already-generated content idea now switches to “生成結果を閉じる” (close result) while the result is shown. * Changed: the top-right “コピー” button now sits at the same height as the open/ close toggle button, and is only visible while the result is open. #### 0.6.4 * Changed: clicking “生成結果を表示” (view result) on an already-generated content idea now toggles the result open/closed instead of always re-fetching and showing it. * Changed: renamed the “今回は不要” idea-card button to “破棄” (discard). * Changed: the “コピー” button on a generated article now appears in both the top- right and bottom-right of the result box (previously only below the content). #### 0.6.3 * Changed: content generation now writes articles as valid HTML (real headings, bold text, tables, lists) instead of Markdown syntax, so pasting the result directly into the WordPress Visual editor reproduces the formatting correctly instead of showing literal `##`/`**`/`|` characters. * Changed: the “コピー” button on a generated article now copies the rendered formatting( rich text), not plain text, so pasted content keeps its heading/bold/table structure in the Visual editor. * Changed: simplified several in-plugin explanation texts (content generation, FAQ generation) to focus on what each feature does for you, removing internal implementation details that aren’t useful for deciding how to use the feature. #### 0.6.2 * Changed: content generation idea cards now span the full width of the admin content area (previously constrained by WordPress admin’s default `.card` max-width). Generated content is now displayed directly beneath the specific card whose “ 生成結果を表示”/”下書きを生成する” button was clicked, instead of in one shared area at the bottom of the page. #### 0.6.1 * Fixed: previously generated content-generation drafts couldn’t be viewed again after reloading the page or navigating away — only ideas whose generation was actively being watched in the same browser tab showed a result. Idea cards for already-generated ideas now show a “生成結果を表示” (view result) button that re-fetches the stored draft. #### 0.6.0 * Changed: the content generation screen has been completely redesigned. Instead of evaluating and rewriting your existing pages, it now analyzes your whole site from a GEO (Generative Engine Optimization) perspective and proposes brand-new content topics the site is missing — informed by your existing content, sampling- investigation citation gaps (Plan A), and your FAQ categories. Pick a proposed idea to generate a full new draft article; the monthly generation-count limit is only consumed when you actually generate an article, not when analyzing the site. * Added: five distinct content angles (direct answer to an uncited search query, comparison/decision-support articles, glossary/definition articles, FAQ-category deep dives, and roundup/hub articles linking related existing posts) that the AI selects automatically based on what your site actually needs. * Removed: the previous “evaluate existing pages generate rewrite” flow. A lighter- weight, generation-free “improve this existing page” feature is being considered separately for a future release. #### 0.5.0 * Changed: the FAQ bulk-generation screen’s AI-assisted mode now runs the “選択した候補を生成する” (generate) step as a background job as well, completing the full site-profiling analysis generation pipeline’s move to the server. You can close the browser tab immediately after starting generation; drafts are created automatically once AI generation finishes. * Added: WordPress draft creation for AI-generated FAQ content is now handled by a background process — triggered when you return to wp-admin, and additionally via a new WP-Cron schedule (every 10 minutes) so drafts are created even if you don’t reopen the admin screen. Duplicate draft creation is prevented even if multiple of these triggers fire around the same time. #### 0.4.1 * Changed: in the FAQ bulk-generation screen’s AI-assisted mode, “サイトを解析する”( site profiling + scoring) now runs as a background job on the server, same as content generation’s evaluate/generate steps. You can close the browser tab immediately after starting, and the ranked candidate list will be waiting when you return. The structural (heading-based, no AI) mode is unaffected and works exactly as before. #### 0.4.0 * Changed: content generation’s “選択した候補を生成する” (generate) step now runs as a background job on the server, same as the evaluate step already did. Previously, generating multiple selected candidates required keeping the browser tab open and active for the entire duration; now you can close the tab immediately after starting and the generated articles will be waiting when you return. * Added: generated content is now cached per page (skipped automatically on re- generation if the source content hasn’t changed since the last successful generation), so re-running generation on unchanged content no longer re-incurs AI cost. #### 0.3.3 * Fixed: reopening the content generation page after evaluation completed with zero candidates (e.g. every item failed due to a temporary AI provider error) showed a blank screen with no way to retry, instead of the “no candidates found” message. This only affected the page-reload state-restore path; the same case was already handled correctly while the evaluation was actively in progress. #### 0.3.2 * Changed: content generation’s “候補を評価する” (evaluate candidates) step now runs as a background job on the server instead of requiring the browser tab to stay open. Uploading the target pages still happens immediately, but the AI evaluation itself continues even after closing the admin screen, and progress/results are restored automatically when you come back. * Added: evaluated results are now cached per page (skipped automatically on re- run if the content hasn’t changed since the last evaluation), so re-running evaluation on an unchanged site no longer re-incurs AI cost. * Added: an internal daily/monthly AI cost limit (admin-configurable, unlimited by default) that stops new AI calls and notifies the administrator once the threshold is reached, resolving automatically once usage drops back under the limit. #### 0.3.1 * Added: a “中断” (Stop) button on both the FAQ bulk-generation box and the content generation page, letting you interrupt the per-page evaluate/analyze or generate loop partway through on large sites. Progress made so far is preserved, and clicking the original button again resumes from where it left off. #### 0.3.0 * Added: content generation support (“コンテンツ生成”). Evaluates existing posts/ pages for GEO improvement opportunities, ranks candidates by priority, and generates a full-length replacement article (with E-E-A-T and PREP structure) for the ones you choose to act on. Results are shown on-screen with a copy button rather than being saved as a draft post. * Changed: renamed “FAQ” menu/list-screen title to “FAQ生成” and the access-analysis page title to drop the redundant “GEO Plugin” / “（プランA）” prefixes, for consistency across the admin menu. * Changed: reordered the admin submenu (FAQ生成コンテンツ生成アクセス解析設定, moving 設定 to the bottom). * Fixed: the top-level menu icon looked dimmer than other menu items when off-hover on newer WordPress versions, and didn’t match the hover text color on older WordPress versions. It’s now tinted to match surrounding menu items on both. #### 0.2.6 * Changed: updated the default backend/update-check URLs to the new platform domain( app.geo-plugin.com), migrated from the previous api.geo.garasaya.com. #### 0.2.5 * Added: an expandable (collapsed by default) explanation of how candidate pages are selected and how the “fewer / balanced / more” count suggestions are derived, shown under “Generation mode: AI analysis” in the FAQ bulk generation box. #### 0.2.4 * Fixed: FAQ content generation used the same 15-second timeout as lightweight requests, causing “cURL error 28: Operation timed out” skips on real sites since it’s the heaviest AI call. Extended to 45 seconds. * Changed: the bulk generation completion message now reads “incorporated: N pages( skipped: M) P FAQ posts” for clarity, and the skipped-item details are now collapsed into an expandable section. * Changed: the “fewer / recommended / more” count buttons no longer show the actual number (to avoid implying it’s user-editable), and “recommended” has been renamed to “balanced”. #### 0.2.3 * Fixed: clicking “Generate FAQs” after reviewing the suggested count for a while could silently do nothing and get stuck showing “Generating: 0 / 0 (0 skipped)”, because the analysis results expire after a period of inactivity. The expiry window has been extended (15 to 60 minutes) and the plugin now shows a clear message asking to re-analyze the site instead of silently freezing. #### 0.2.2 * Fixed: the “recommended” and “more” FAQ generation count suggestions could collapse to the same number when the AI’s suggested baseline was close to or above the number of usable candidates, defeating the purpose of offering distinct choices.“ Recommended” now reserves headroom so “more” stays meaningfully higher whenever there are enough candidates to support it. #### 0.2.1 * Fixed: the actual update ZIP download (as opposed to the update-check request already fixed in 0.1.2) could still fail with “cURL error 60” on outdated-hosting environments, because WordPress core downloads the package itself via a separate HTTP request that bypassed the plugin’s bundled CA certificate. It now applies to this request too, scoped to this plugin’s own distribution URL only. * Fixed: the dynamic FAQ generation count labels (“fewer/recommended/more”) showed the literal placeholder text instead of the actual numbers. * Added: the FAQ candidate analysis list is now collapsed into an expandable section by default. #### 0.2.0 * Added: FAQ bulk generation (AI-enhanced mode) now analyzes the whole site once per run to understand the business and propose site-specific FAQ categories, and consolidates each category’s Q&A into a single draft post instead of one post per source page. * Added: generation count is now suggested dynamically (“fewer” / “recommended”/“ more”) based on the site analysis, instead of a fixed list. The unbounded “all” option has been removed. * Changed: per-page relevance scoring now uses a longer excerpt and site-wide context for better judgment on less-structured sites. #### 0.1.2 * Fixed: the update checker itself (used to detect new plugin versions) could still fail with “cURL error 60: unable to get local issuer certificate” on the same outdated-hosting environments addressed in 0.1.1, because it communicates independently of the backend API client. It now uses the same bundled CA certificate file. #### 0.1.1 * Fixed: backend communication could fail with “cURL error 60: unable to get local issuer certificate” on hosting environments where the WordPress-bundled CA certificate file is outdated. The plugin now bundles and uses its own current CA certificate file for backend requests. #### 0.1.0 * Initial release * Automatic Schema.org JSON-LD output (Article, Organization, WebSite, FAQPage, BreadcrumbList) * robots.txt / sitemap.xml / llms.txt management * FAQ generation (structural analysis and AI-assisted) * AI-enhanced schema properties (description, abstract, keywords) with auto-trigger on publish * BreadcrumbList shortcode ## Meta * Version **1.0.0** * Last updated **1 day ago** * Active installations **Fewer than 10** * WordPress version ** 4.9 or higher ** * Tested up to **7.1.1** * PHP version ** 7.0 or higher ** * Tags * [AI](https://wordpress.org/plugins/tags/ai/)[faq](https://wordpress.org/plugins/tags/faq/) [geo](https://wordpress.org/plugins/tags/geo/)[schema](https://wordpress.org/plugins/tags/schema/) [seo](https://wordpress.org/plugins/tags/seo/) * [Advanced View](https://wordpress.org/plugins/garasaya-aio-dash/advanced/) ## Ratings No reviews have been submitted yet. [Your review](https://wordpress.org/support/plugin/garasaya-aio-dash/reviews/#new-post) [See all reviews](https://wordpress.org/support/plugin/garasaya-aio-dash/reviews/) ## Contributors * [ garasaya ](https://profiles.wordpress.org/garasaya/) ## Support Got something to say? Need help? [View support forum](https://wordpress.org/support/plugin/garasaya-aio-dash/)