外掛說明
AI 外掛將 AI 功能直接整合至 WordPress 管理及編輯體驗中。
需要使用 WordPress 區塊編輯器。不支援 Classic Editor 外掛及其他非區塊編輯器的編輯體驗。
內含功能:
這個外掛以 AI Building Blocks for WordPress 計畫為基礎,整合 AI Client 程式庫與 Abilities API,提供一致的使用體驗。它既是內容創作者的實用工具,也是開發者可參考的實作。
現有功能:
- 能力總管 – 可在專屬管理畫面瀏覽及操作已註冊的 AI 能力。
- AI 請求記錄 – 記錄 AI 請求,以提供可觀測性及偵錯資訊。
- 替代文字產生 – 為圖片產生具描述性的替代文字,以改善無障礙功能。
- 留言審核 – 根據毒性偵測與情緒分析,自動審核留言。
- AI Connector 核准 – 外掛或佈景主題必須先取得網站管理員明確核准,才能使用這個網站已設定的 AI Connector。
- 內容分類 – 建議相關的標籤與分類,以利整理內容。
- 內容長度調整 – 縮短、擴充或改寫選取的區塊內容。
- 內容摘要 – 將長篇內容濃縮成便於掌握的摘要。
- 內容翻譯 – 可在文章編輯器中,將段落及標題區塊翻譯為選取的語言,並可選擇是否翻譯文章標題。
- 自訂能力 – 將外掛的通用 WordPress 能力集中由單一開關控制是否啟用。
- 控制台小工具 – AI 狀態及 AI 能力小工具,並提供可註冊新小工具的架構。
- 編輯備註 – 逐一檢閱文章內容區塊,並加入備註,提供無障礙、可讀性、文法及 SEO 改善建議。
- 編輯內容更新 – 自動將編輯備註套用至內容。
- 內容摘要產生 – 自動為文章建立精簡摘要。
- 實驗架構 – 可選擇加入的機制,讓你只啟用想使用的 AI 功能。
- 準則 – 讓能力遵循全站的編輯標準。
- 圖片產生與編輯 – 可在編輯器中根據文章內容產生及編輯圖片,也可透過媒體庫操作。
- 金鑰加密 – 使用內附的 libsodium 加密功能,將靜態儲存的 AI 供應商 API 金鑰加密。讀取時會自動解密,寫入時則會重新加密。停用此實驗功能或停用外掛後,金鑰會還原為純文字。
- 中繼描述產生 – 產生中繼描述建議,並將建議整合至各種 SEO 外掛。
- 多供應商支援 – 可搭配 OpenAI、Google 與 Anthropic 等供應商的 AI Connector 外掛使用。
- 代稱產生 – 在永久連結彈出式視窗或發佈前面板中,取得適合 SEO 的文章永久連結代稱建議,再編輯並套用想使用的項目。
- 建議回覆 – 在 [留言] 畫面及 [活動] 小工具加入「建議回覆」操作,讓審核人員能快速產生留言回覆建議。
- 標題產生 – 按一下即可為文章產生標題建議。非常適合用來發想標題,或為內容找到合適的語氣。
- 自動完成 – 輸入時提供符合情境的即時建議。
供應商設定:
AI 外掛本身不含供應商憑證或供應商實作。若要使用 AI 功能,請至少安裝並啟用一個 AI Connector 外掛,再前往 Settings -> Connectors 設定其憑證。在 AI Connector 完成安裝及驗證,並能執行所需操作前,相關功能可能會顯示為無法使用。
供應商連接器外掛包括 Anthropic、Google、OpenAI 及其他外掛。
即將推出:
我們正積極開發新功能,以強化你的 WordPress 工作流程:
- AI 試驗場 – 試用不同的 AI 模型及供應商。
- 內容助理 – Gutenberg 中由 AI 輔助的寫作及編輯功能。
- 網站代理 – 使用自然語言管理 WordPress。
- 工作流程自動化 – 由 AI 驅動的工作自動化。
這是實驗性外掛;隨著我們蒐集社群意見,功能可能會有所變更。
發展藍圖:
你可以在 WordPress AI GitHub 專案看板的篩選檢視中,查看外掛目前的開發藍圖。
開發者專區
AI 外掛的設計可供研究、擴充,並在其基礎上進一步開發。無論你是外掛開發者、代理商或主機服務供應商,都可以進行以下工作:
擴充外掛:
- 建立自訂實驗功能 – 使用
Abstract_Feature基底類別,建立自己的 AI 功能。 - 預先設定供應商 – 主機服務商及代理商可設定 AI Connector 外掛,讓使用者不必自行準備 API 金鑰。
- 能力總管 – 測試及探索已註冊的 AI 能力(啟用實驗功能後可用)。
- 註冊自訂能力 – 串接 Abilities API 以加入新的 AI 功能。
- 覆寫預設行為 – 使用篩選器自訂提示、回應及使用者介面元素。
- 完整的 hook 支援 – 整個程式碼庫均提供可供自訂使用的篩選器及動作。
即將推出的開發者工具:
- AI 試驗場 – 試用不同的 AI 模型與提示詞。
- MCP (Model Context Protocol) – 在 WordPress 工作流程中整合並測試 Model Context Protocol 的功能。
- 擴充供應商 – 支援以額外或替代的 AI 供應商進行實驗。
開始使用:
- 閱讀參與貢獻指南,瞭解如何設定開發環境
- 前往 WordPress Slack 的 #core-ai 參與討論
- 瀏覽 GitHub 儲存庫,瞭解實驗功能的建置方式
- 參與討論,交流外掛後續該如何改進。
歡迎參與貢獻!無論你想建立新的實驗、改善現有功能,或協助撰寫文件,都可前往我們的 GitHub 儲存庫參與。
螢幕擷圖
![功能:圖片產生與編輯。文章編輯器側邊欄顯示 [產生精選圖片] 按鈕,以及已產生的精選圖片預覽,並包含替代文字、標題及內容說明。](https://ps.w.org/ai/assets/screenshot-1.gif?rev=3607988)
![編輯器實驗功能:替代文字產生。圖片區塊設定顯示 [產生替代文字] 按鈕及產生的替代文字。](https://ps.w.org/ai/assets/screenshot-4.gif?rev=3607988)


![編輯器實驗功能:留言審核。[活動] 小工具的 [近期留言] 區段顯示情緒及毒性分數。](https://ps.w.org/ai/assets/screenshot-7.png?rev=3607988)

![編輯器實驗功能:內容摘要。文章編輯器側邊欄顯示 [產生 AI 摘要] 按鈕,以及 [內容摘要] 區塊中產生的內容摘要。](https://ps.w.org/ai/assets/screenshot-10.gif?rev=3607988)
![編輯器實驗功能:內容摘要產生。文章編輯器側邊欄顯示 [產生內容摘要] 按鈕及產生的內容摘要。](https://ps.w.org/ai/assets/screenshot-13.gif?rev=3607988)

![編輯器實驗功能:標題產生。文章編輯器在文章標題欄位上方顯示 [產生] 按鈕,並在互動視窗中顯示標題建議。](https://ps.w.org/ai/assets/screenshot-15.gif?rev=3607988)







![管理介面實驗功能:AI 請求記錄。記錄 AI 請求,以供可觀測性及偵錯使用。可在 [工具] 中檢視詳細記錄。](https://ps.w.org/ai/assets/screenshot-23.png?rev=3607988)


![管理介面實驗功能:建議回覆。在 [留言] 畫面及 [活動] 小工具加入「建議回覆」操作,讓審核人員能快速產生留言回覆建議。](https://ps.w.org/ai/assets/screenshot-26.gif?rev=3653536)

安裝方式
- 將外掛檔案上傳到
/wp-content/plugins/ai目錄,或直接透過 WordPress 外掛畫面安裝外掛。 - 在 WordPress 的 [外掛] 畫面啟用這個外掛。
- 安裝並啟用至少一個 AI Connector 外掛,接著前往
Settings -> Connectors設定其認證資訊。 - 前往
Settings -> AI,先全域啟用功能,再啟用想測試的個別功能或實驗。 - 開始試用各項 AI 功能!若要使用 [標題產生] 實驗功能,請編輯文章並點進標題欄位。欄位上方應會顯示
Generate/Regenerate按鈕。按一下該按鈕,請求完成後,標題建議會顯示在互動視窗中。選擇喜歡的標題,再按一下Select按鈕,即可將其插入標題欄位。
常見問題集
-
這個外掛有什麼用途?
-
這個外掛將 AI 寫作及編輯工具直接整合至 WordPress,同時也提供參考實作,供想要建立自有 AI 功能的開發人員使用。
-
在正式網站使用這個外掛安全嗎?
-
這是實驗性外掛,因此建議先在測試環境中試用。隨著我們蒐集社群意見,功能可能有所變動。所有 AI 功能都需由使用者選擇啟用並手動觸發;未經你的核准,不會自動執行任何操作。
-
支援哪些 AI 供應商?
-
這個外掛可搭配 Anthropic(Claude)、Google(Gemini)、OpenAI 及其他供應商的 AI Connector 外掛使用。請安裝並啟用所需的 AI Connector 外掛,再前往
Settings -> Connectors完成設定。 -
需要 API 金鑰才能使用這些功能嗎?
-
是,目前必須為已設定的 AI Connector 外掛提供自己的 API 金鑰,例如 OpenAI、Google AI 或 Anthropic。
-
費用是多少?
-
外掛本身免費,但仍需支付所選 AI 供應商的 API 用量費用。費用因供應商及用量而異。多數供應商會提供免費試用額度供你開始使用。也可使用部分在本機執行、採用開放原始碼且免費的供應商(例如 Ollama)。
-
沒有程式設計知識也能使用嗎?
-
當然可以!這個外掛專為內容創作者和網站管理員而設計。設定好 AI Connector 後,就能直接在文章編輯器中使用 AI 功能。
-
這個外掛支援 Classic Editor 嗎?
-
不支援。
AI 外掛目前僅支援區塊編輯器(亦稱 Gutenberg)。此外掛的設計以現代化的編輯器 API、區塊式內容工作流程,以及 WordPress 核心(包括 Gutenberg)持續發展中的編輯功能為基礎。
自 WordPress 5.0 於 2018 年推出以來,區塊編輯器一直是 WordPress 的預設編輯體驗,至今仍是編輯器積極開發的主要焦點。將開發資源集中在區塊編輯器,能讓此專案更快推出新功能、實驗及整合,同時避免在多種編輯體驗中維護平行實作的複雜性。
雖然 Classic Editor 外掛仍有相當龐大的安裝量,Core AI 團隊仍決定優先在區塊編輯器生態系中推動創新及實驗。目前沒有新增 Classic Editor 官方支援的計畫。
-
哪裡可以取得協助或回報問題?
-
你可以在 WordPress Slack 的 #core-ai 頻道提問,或到 GitHub 儲存庫回報問題。
使用者評論
參與者及開發者
變更記錄
1.3.0 – 2026-08-18
Added
- New Experiment: Content Translation; translates Paragraph and Heading blocks—and optionally the post title—into a selected language directly from the post editor (#747).
- New Experiment: Slug Generation; suggest SEO-friendly permalinks that can be set as the slug (#897, #932).
- New Experiment: Custom Abilities. Gates the plugin’s custom WordPress Abilities (
ai/get-post-details,ai/get-post-terms,core/read-settings,core/read-users,core/read-content) behind a single opt-in toggle, so enabling it exposes all of them at once via the Abilities API. Note for anyone that is using these Abilities, you’ll need to enable this new experiment first for those to be available (#881). - New Developer Tool: Import/Export functionality for non-sensitive AI settings (#734).
- Cleanup plugin data when the plugin is deleted (#692).
- AI-specific Site Health integration and status tests (#734).
- New filters,
wpai_content_classification_available_terms,wpai_content_classification_min_confidenceandwpai_content_classification_candidate_pool_size, to allow more control over Content Classification (#633). - Prompt template extension points, making it easy for others to filter prompts and prompt builders (#770).
- Brought the embedding code over from the PHP AI Client and load that conditionally so those using the AI plugin can start to take advantage of embedding generation (#892, #946).
- Public
WordPress\AI\log_ai_request()API so MCP servers and ability consumers can record requests in the AI Request Log (#914).
Changed
- Updated all meta keys to use the
wpai_prefix instead of justai_. Note this changes the prefix on theai_generated,ai_generated_summaryandai_notemeta so if you are directly using those, update to using thewpai_names (#867). - Updated preferred models to more recent ones for the three default providers (#913).
- Bump WordPress tested-up-to version 7.1 (#934).
- Improve the relevance of category and tag suggestions produced by the
Content Classification experiment (#633). - Editorial Updates now links to the visual revisions screen when reviewing refined content, falling back to the classic revisions screen when visual revisions are unavailable (#861).
- Reordered setting experiments list; grouped linked experiments and sorted editor experiments alphabetically (#871).
- Improved keyboard focus handling when generating, accepting, or dismissing classification suggestions (#873).
- The Abilities Explorer provider filter dropdown now includes custom providers, and the overview statistics count abilities by origin so custom-provider abilities remain in their Core/Plugins/Theme bucket (#884).
- Set focus to the generated title textarea when generating a title (#901).
- The
core/read-usersability now returns collections ordered by display name, A to Z (#948).
Deprecated
- The
AI_Serviceclass and theget_ai_service()helper introduced in 0.2.1 will be removed in the next major release. Neither is used anywhere in the plugin; experiments and abilities callwp_ai_client_prompt()directly (#905). - Filter
wpai_meta_description_result_temperatureis no longer being used and will be removed in the next release (#913).
Removed
- No longer set custom temperature values on any of our requests (#913).
Fixed
- The AI Request Log REST endpoint now registers its
operationfilter parameter, so it appears in the REST schema and a non-string value returns a 400 instead of causing a fatal error (#758). - Inline reply textarea not receiving focus after generating a suggested reply (#877).
- Meta Description suggestions applied on pages and custom post types were lost on save when Yoast SEO was active (#886).
- Improved accessibility and keyboard usability for the request logs provider/model details (#889).
- Improved keyboard and focus handling for the Suggest Reply tone dropdown (#907).
- Synchronized generating state across the inline and modal excerpt generation buttons (#908).
- Ensure caller detection in the encryption experiment properly matches the calling plugin, not the host plugin (#909).
- Synchronized loading state between the sidebar and block toolbar regenerate summary buttons (#912).
- Preserve inline HTML when resizing content (#915).
- Bulk actions no longer re-run when sorting or paginating the list after the action completes (#928).
- Apply editorial updates to blocks that store editable text in the
valueattribute (#930).
Security
- Ensure any content we render from the LLM or content we send to the LLM is properly sanitized (#950).
- Add proper nonce check prior to bulk alt text and summarization generation (GHSA-hfp9-55vw-ccjc).
- When passing a custom image URL to the Alt Text Generation Ability, ensure that URL is public, that it points to an allowed image type and that the final URL we download matches the initial one we verify (GHSA-v2wx-9j88-4rqq).
1.2.0 – 2026-07-14
Added
- New Experiment: Suggest Reply; gives comment moderators a quick way to generate a reply to a comment through the admin (#724).
- New “Advanced settings” option in Developer Tools to show/hide additional configuration options for features and experiments (#842).
- Bulk “Generate AI Summary” action to the posts and pages list table, enabling summary generation for multiple posts at once (#650).
- New
core/read-contentAbility with secure single-post and query modes (including include and opt-in fields) to enable read-only content access (#739). - New
core/read-usersAbility that retrieves a single user by ID, email, login, or nicename, or a filtered and paginated users collection, with sensitive fields opt-in and permission-gated (#774). - An inline admin notice when Connector Approvals are enabled and no AI connectors are yet approved, prompting admins to approve the AI plugin for use (#830).
- Introduce new
wp_ai_client_default_request_timeoutfilter to make the default request timeout configurable. Use this for the image generation request timeout (#862).
Changed
- Content Summary block detection now checks within nested blocks (#810).
- Move all existing configuration options into a new “Advanced settings” section which is hidden by default (#842).
Fixed
- Focus restoration after AI setting saves (#812).
- Added descriptive alt text to AI Home feature card images for improved screen reader accessibility (#819).
- Prevent Type-Ahead assets from loading on the front end (#820).
- Dismissing a type-ahead suggestion with escape should not trigger a new suggestion request (#840).
- Type-ahead ghost text placement and stale suggestions overlapping empty-block placeholders (#847).
MutationObservercrash when editor iframe body isn’t ready when using Title Generation (#849).- Respect an explicit
show_in_abilitiesvalue on curated settings, and leave the flag to WordPress core once core declares it (#852). - Register initial settings before
core/read-settingssnapshots them (#856).
1.1.0 – 2026-06-30
Added
- New Experiment: Type Ahead; automatically suggests ghost text at the end of paragraphs, can be manually triggered within a paragraph (#151, #776).
- New Experiment: Key Encryption; encrypts AI Connector API keys before storing them in the database (#560).
- Ensure all Features that rely on utilizing content are disabled until minimum content thresholds are met (#581).
- New
core/read-settingsAbility (#691, #806). - New
wpai_has_image_generation_supportfilter that allows 3rd parties to claim support for Image Generation, for example if authenticating without an API key (#748). - New setting to choose if guest comments should be moderated or not, defaulting to
yes(#751). - Explicit save button to developer settings panel, requiring a user to click save before the Provider and Model settings are saved (#761).
- Documentation callouts that the plugin has targeted support for the Block Editor only (#766).
Changed
- Note prompting the user to save after running Editorial Notes (#682).
- Use
__next40pxDefaultSizefor buttons consistently (#702). - Skip Comment Analysis and Moderation when comment has already been flagged as spam/trash (#743).
- Replace developer mode settings CSS with Stack component (#785).
- Use character-based count instead of word-based to determine when features are available to use (#802).
- Use
Noticecomponent to display warnings within experiment modals (#803). - Added success snackbar when saving or resetting developer settings (#807).
Fixed
- Add
wordCountTypeto check for user’s locale and update to count character or words when detecting minimum content length (#581). - Restrict Content Resizing to REST-exposed post types when a post ID is provided (#658).
- Only show Editorial Updates button when pending Notes are linked to current blocks (#682).
- Hide developer settings on stable Features when AI is disabled (#737).
- Improve readability of Ability Explorer schema output by preventing unicode escaping for non‑ASCII characters (#740).
- The “Last 30 Days” summary period in the AI Requests Logs page now uses a fixed 30-day window so the summary cards and logs table cover the same span (#753).
- Developer Tools popover overlapping the WP admin bar (#756).
- Persistence of suggested terms when running the Content Classification experiment (#769).
- Restored term suggestion pills to their original positions if the backend term assignment API fails, resolving stale state race conditions (#772).
- Preserve omitted
runAbility()input so ability schema defaults can apply when abilities are invoked without input (#775). - Ensure scalar input schemas are allowed in the Abilities Explorer validation (#787).
- Standardize the Title Generation button text (#790).
- Scope the Editorial Note generation loading spinner only to the block currently being reviewed (#794).
- Snackbar notifications no longer overlap the settings content; they are pinned to the bottom-left of the content area (#801).
1.0.2 – 2026-06-15
Added
- Manual refresh button to the AI Request Logs table header (#687).
- New
ai_generatedparam on our Image Import Ability to set if the imported image was AI generated or not (GHSA-42mg-ffvx-4xff).
Changed
- Ensure Editorial Notes and Editorial Updates controls stay grouped together in the post editor sidebar (#605).
- Use explicit UTF-8 encoding for generated Meta Description character counts (#655).
- Return a consistent decorative flag from Alt Text Generation results (#659).
- Show an error message immediately in the Image Generation UI when there’s no AI Connector in place that supports image generation (#679).
- Use a neutral icon for disabled Features and Experiments in the AI Status widget (#720).
Fixed
- Abilities Explorer schema validation (#612).
- Alt Text Generation button becomes unresponsive after using Next/Previous in the media modal (#631).
- Add descriptive accessible labels to approval matrix toggle controls (#637).
- Added accessible labels to the Provider and Category filter dropdowns on the Abilities Explorer page (#642).
- Lost focus after generating a Title (#644).
- Lost focus when generating Alt Text in Image block inspector controls (#645).
- Lost focus when toggling the Connector Approval state (#646).
- Lost focus after generating Images (#647).
- Added an accessible label to the ability test payload textarea in the Abilities Explorer (#649).
- Excerpt generation post context payload (#651).
- Clear out the Meta Description suggestion when the modal closes (#653).
- Lost focus after running Content Resizing actions (#663).
- Column reordering and hiding in the AI Request Logs table now persists instead of resetting to the default (#669).
- Summary statistics showing zero for short time periods on non-UTC MySQL servers (#671).
- UI inconsistency on AI Request Logs page (#676).
- Ensure thinking tokens are counted in AI Request Logs (#680).
- Ensure the Ability schemas and outputs are valid JSON Schema for strict REST and MCP consumers (#688).
- Title Generation button disappears after toggling off “Show template” (#694).
- Prevent accidental interactions and stale feedback in the Meta Description Generation modal and improve focus handling (#696).
- Ensure focus isn’t lost after generating an Excerpt inline (#698).
- AI Request Logs: “Copy Log ID” gives no feedback when copied (#700).
- AI Request Logs: main header overlapping table header (#705).
- Allow users to clear an applied Meta Description while preventing whitespace-only descriptions (#706).
- Rename unforwarded
MaskCanvascomponent function toInnerMaskCanvasto avoid duplicate declarations (#713).
Security
- Remove the
metaparam from our Image Import Ability (GHSA-42mg-ffvx-4xff). - Check the current user’s capabilities and the comment type before setting an Editorial Note (GHSA-j7hg-vqpw-f98f).
1.0.1 – 2026-05-27
Added
- New helper functions that are used to determine if we have valid AI Connector credentials (#603).
- New helper methods,
is_globally_enabledandis_individually_enabledto help tell if a feature is enabled individually or if features are globally enabled (#604).
Changed
- Removed the description from the Abilities listing within the Abilities Explorer (#592).
- Filter Guideline queries by the guideline type content (#593).
- Use the new
has_connector_authenticationinstead ofis_connector_configuredto avoid unnecessary API requests (#603).
Removed
- Deprecated
__nextHasNoMarginBottomprop (#609).
Fixed
- Utilize a new
is_connector_configuredfunction to properly determine if a connector is configured, whether via an API key, constant or ENV var (#537). - “Generate Editorial Note” button appearing in the block settings menu during post revisions (#591).
- If the Connector Approvals experiment is turned on, ensure we don’t over-aggressively block functionality in the AI plugin that isn’t actually making requests, like Request Logging (#595).
- Better matching of the originating code when the Connector Approvals experiment is on (#595).
- Focus loss issues when interacting with Purge actions in the Request Logs experiments page (#599).
- Disable the “Purge All” button when no logs are available to purge (#599).
- AI Status feature checklist properly shows if an individual feature is enabled even if globally features are disabled (#604).
- Ensure focus isn’t lost when buttons enter disabled state during Alt Text Generation, Content Classification, Content Summarization, Excerpt Generation, Featured Image Generation, and Title Generation (#608, #611).
- Settings page strings, which are enqueued as script modules, are now localized at runtime (#613).
- Connector Approvals “Dismiss” button failing for pending requests whose key contains a slash (#615).
- Hide empty provider capabilities section in the dashboard widget (#616).
- Playground and test configs now target the latest WordPress release instead of the beta release (#626).
- Connector Approvals notice no longer overlaps the page header on the AI Request Logs screen (#628).
1.0.0 – 2026-05-19
Added
- New Experiment: Request Logging that provides observability for all AI operations (#437).
- New Experiment: Connector Approvals that allows administrators the ability to determine which plugins can access which AI connectors (#467).
- Integrate Alt Text generation into the experimental media editor (#446).
- Sorting and filtering in Comments screen by Toxicity and/or Sentiment (#518).
- Toxicity and Sentiment labelling in admin dashboard for comments (#518).
Changed
- Disable the Summarization button until content reaches a certain length (#492).
- Refined image generation loading state (#512).
- Featured image button now hides when image is already set (#512).
- When no AI provider is configured and a feature is triggered, show actionable guidance directing users to configure an AI Connector (#523).
- Update Meta Description loading state and remove duplicate heading in modal (#527).
- Rename “Review Notes” experiment to “Editorial Notes” and “Refine from Notes” experiment to “Editorial Updates” (#528).
- Keep comments without moderation metadata visible when sorting by Comment Moderation columns (#538).
- Updated plugin banner and icons (#546).
- Show a notice when a user has chosen a provider that no longer exists (#552).
- When no provider is configured, show an error notice instead of an admin notice for alt text generation (#561).
- Standardize error message text (#562).
- Abilities Explorer page heading (#585).
Fixed
- Ensure we properly use the new client-side Abilities API (#482).
- Keep keyboard focus on the Provider select when resetting per-feature developer settings to default (#532).
- Deduplicate provider API requests on the settings page when developer mode is toggled on (#542).
- Update the Playground Preview workflow to use
pluginDatainstead ofpluginZipFile(#548). - Empty space shown for Model field when saved provider no longer exists in developer settings (#552).
- Prevent analyzing newly inserted comments when no provider is configured (#554).
- Ensure the meta description modal doesn’t open if no provider is configured (#558).
- False error for alt text generation on decorative images in media library (#559).
- Show a failed badge when comment analysis fails (#568).
- Correct RTL rendering of directional icons, runtime-set styles, and inline styles in the admin UI (#573).
- Add notice to standalone image generation when there is no provider connected (#575).
- Ensure we show a more specific error message when no valid AI connector is in place and we try to generate a featured image (#576).
- Improve keyboard focus visibility for suggested term actions in content classification (#580).
- User-facing text in several experiments is now fully translatable, and JS-side translations are loaded at runtime (#582).
- Make title generation and content classification UI react to current editor state (#584).
- Ensure global AI enabled options are migrated properly (#586).
Older changelog entries can be found in the CHANGELOG.md file.
