Sentient AI - WordPress AI Chatbot with Local Brain, Autopilot & Empathy Engine -- WordPress 插件

开发者	maidulcu
更新时间	2026年9月11日 18:55
PHP版本:	7.4 及以上
WordPress版本:	7.1
版权:	GPLv2 or later
版权网址:	版权信息

2.45.0

New: System Info page — New admin page showing PHP version, WordPress version, memory limits, server details, and plugin status for debugging. Copy to Clipboard button for easy sharing in support tickets
Fix: EscapeOutput escaping — All badge outputs now use wp_kses_post() for PHPCS compliance (WordPress.Security.EscapeOutput.OutputNotEscaped)
Fix: Copy to Clipboard — Replaced navigator.clipboard API with execCommand fallback for HTTP compatibility and broader browser support

2.44.0

New: RTL support (Multilingual Part 2) — Full right-to-left layout for Arabic, Hebrew, Persian, and Urdu: mirrored chat bubbles, right-aligned input, flipped list indentation; a language badge now appears on bot messages
New: FAQ/KB response localization (Multilingual Part 3) — Matched FAQ and semantic-vector answers are auto-translated into the visitor's language; translation results cached for 30 minutes
Fix: Language instruction on Hugging Face — Language instruction now also embedded in the user query (not only the system prompt) so open-source models like DeepSeek respond in the selected language
Fix: FAQ/KB priority — When knowledge base context is found, the redundant FAQ translation API call is skipped, reducing credit usage and returning richer answers
Fix: Available Languages reset — Language settings are now preserved when saving from any settings tab, so the configured languages no longer reset to just "en"
Revert: max_tokens — Restored hardcoded token limits (OpenRouter 300, OpenAI 500) to avoid OpenRouter credit limit errors

2.43.0

New: Multilingual Chatbot — Automatically detect each visitor's browser language and instruct the AI to respond in the same language
New: Language Handler — New Language_Handler class with support for 30+ ISO 639-1 languages (English, Arabic, Spanish, French, German, Hindi, Chinese, etc.), browser Accept-Language detection, and Unicode-script text detection
New: Language switcher — A language dropdown appears in the chat header when multilingual mode is enabled, letting visitors switch languages manually
New: Admin settings — New "Multilingual Chatbot" card on the Features tab with enable toggle, default language dropdown, and configurable available languages list
New: Language-aware system prompt — The AI system prompt is augmented with a language instruction (e.g. "Respond in العربية (ar)") when multilingual mode is on and a non-English language is detected
Note: Privacy Mode — Chatbot answers only from local FAQs when Privacy Mode is enabled; multilingual AI responses require Privacy Mode to be off

2.42.0

New: Website Scraping — Scrape public webpages by URL and add them to the knowledge base; extracts readable text from HTML pages (strips scripts, styles, SVGs), reuses the same KB chunking and embedding pipeline
New: KB search answers via AI (RAG) — When the chatbot finds relevant knowledge base content, it now passes the full chunk text to the AI as context for a detailed, structured answer, instead of returning a raw 200-character preview
New: Markdown rendering in chatbot — Bot responses now properly render markdown formatting: **bold**, ## headings, - bullet lists, `code`, fenced code blocks, and horizontal rules
Fix: Web Scraper regex — Fixed invalid PCRE2 Unicode range syntax (\u0600 → \x{0600}) that caused the scraper to silently fail on all URLs
Fix: Web Scraper HTML extraction — No longer strips <header> tags which commonly contain main page content (site titles, descriptions, hero sections)
Fix: Chatbot HTML entity display — Fixed double-escaping bug where & appeared as literal text in bot responses

2.41.0

New: Multi-provider embeddings — KB and post vector search now supports OpenAI, OpenRouter, Google Gemini, and Hugging Face embedding providers, with an "Embedding Provider" selector in Tools → Vector DB
New: Auto-detect embedding provider — auto mode automatically uses the first provider with a configured API key (OpenAI → OpenRouter → Gemini → Hugging Face → local fallback), so semantic search works without an OpenAI key
New: Per-provider embedding models — Configurable embedding model per provider (OpenRouter, Gemini, Hugging Face) with sensible defaults
Fix: Embedding dimension mismatch — cosine_similarity() now returns 0 for vectors with mismatched dimensions instead of silently truncating, preventing garbage scores when providers are switched
Fix: Knowledge Base source icon — Source attribution SVG icons now render correctly in the chatbot (was displayed as raw HTML); added a dedicated "Knowledge Base" source type with a book icon and label

2.40.0

New: Knowledge Base Vector Search — Uploaded documents can now be indexed for semantic search: each chunk gets an AI embedding (1536-dim) stored in the sentient_ai_kb_chunks table; chatbot queries now match against document chunks and return them with source attribution
New: "Index for Search" button — Files in processed state show an index button on the Training page; one click generates embeddings for all chunks and moves the file to indexed status
Fix: Embedding provider logic — generate_embedding() now uses OpenAI embeddings whenever an OpenAI API key is configured, independent of the chat provider, fixing a dimension-mismatch bug (128-dim fallback vs 1536-dim real) that silently disabled vector search for non-OpenAI chat providers
Fix: KB search on frontend — KB_Handler is now loaded on both frontend and admin so the chatbot can search knowledge base chunks from visitor-facing pages
Update: WordPress 7.1 tested — Tested up to bumped from 6.9 to 7.1
Docs: KB embeddings plan — Added KB_EMBEDDINGS_PLAN.md documenting the chosen future approach (Option B: dedicated Embedding Provider setting) for OpenRouter/Gemini/Hugging Face embedding support

2.39.0

Fix: Frontend fatal error — Wrapped KB_Handler::init() in class_exists() guard to prevent "Class not found" fatal error on frontend pages where the KB file was not loaded

2.38.0

New: Document Chunking — Uploaded documents are automatically split into ~1000-char chunks with 150-char overlap for vector-based retrieval, stored in new sentient_ai_kb_chunks table
New: KB File Processing & Storage — Files are persisted to disk (wp-content/uploads/sentientai-kb/) with metadata in sentient_ai_kb_files table; status flow: pending → processed → indexed
Fix: Full text preservation — Complete document text is now saved instead of only the first 4000 characters; uses server-side transient token bridge to pass full text from parse to save step
Fix: Chunking logic for single-paragraph documents — Large documents without paragraph breaks are now properly split into multiple chunks instead of being stored as one giant chunk
Fix: Duplicate chunk count badge — File list now correctly shows chunk count after AJAX save, not just on page reload

2.37.0

New: Knowledge Base File Upload — Upload TXT, MD, DOCX, and PDF documents on the Training page with drag & drop or click-to-browse, plus instant extracted-text preview with word count
New: File Parser Engine — New File_Parser class extracts plain text from PDFs (pdftotext binary or internal FlateDecode stream extraction), DOCX (ZIP/XML parsing), and plain-text files (TXT/MD/CSV) with automatic encoding conversion
Security: Strict file type verification — Magic-byte checks and control-character content sniffing reject renamed binaries (e.g. images disguised as .pdf); extension whitelist and 5 MB size limit enforced on both client and server side

2.36.0

New: Topic Frequency Bar Chart — Horizontal bar chart on Learning page showing most frequent topics with 7/30/90 day period selector, replacing the plain Top Topics table
New: Most Matched FAQs Chart — Horizontal bar chart on Query Logs page showing which FAQs are matched most frequently with 7/30/90 day period selector
Fix: Learning tables auto-creation — Added ensure_tables_exist() to Memory_Handler so learning, topics, and answer quality tables are created automatically on upgrade

2.35.0

New: Sentiment Over Time Chart — Multi-line SVG chart on Sentiment page showing daily trends for happy, neutral, frustrated, confused, and urgent emotions with 7/30/90 day period selector
New: Sentiment Timeline AJAX Handler — New endpoint returns daily emotion counts grouped by date and sentiment
Fix: Sentiment chart tooltips — Tooltips now appear on hover by fixing innerHTML re-render that was destroying event listeners
Fix: Chart period button sync — Emotion Distribution and Sentiment Over Time chart buttons now reflect the global Time Range selection on page load

2.34.0

New: Emotion Distribution Chart — SVG donut pie chart on Sentiment page showing breakdown of happy, neutral, frustrated, confused, and urgent emotions with 7/30/90 day period selector
Fix: WordPress 7.1 iframed editor compatibility — Schema meta box SERP preview now uses wp.data store for block editor, with classic editor fallback
Fix: FAQ block script dependency — Removed unnecessary wp-editor dependency from FAQ Gutenberg block
Update: Prepared for WordPress 7.1 — Verified code compatibility with enforced iframed editor, jQuery UI 1.14.2, block CSS loading changes, and accessibility improvements. Full audit: no deprecated functions, no wp_editor() usage, no jQuery UI widget conflicts, proper block.json registration.

2.33.0

New: Top 10 Queries Bar Chart — Horizontal bar chart on Query Logs page showing most frequent queries with 7/30/90 day period selector
Update: Query Logs Layout — FAQ vs AI pie chart and Top Queries bar chart now display side by side in a 2-column grid
Fix: FAQ vs AI Chart initialization — Pie chart now loads automatically on page reload

2.32.0

New: FAQ vs AI Response Ratio Chart — SVG donut pie chart on Query Logs page showing FAQ vs AI response percentage with 7/30/90 day period selector
New: FAQ vs AI Chart AJAX Handler — New endpoint returns FAQ and AI response counts with percentages
Update: Pie Chart Design — Enhanced visual design with gradient progress bars, drop shadows, and improved legend layout

2.31.0

New: Queries Over Time Chart — SVG line chart on Query Logs page showing daily query count with 7/30/90 day period selector
New: Query Chart AJAX Handler — New endpoint returns daily query counts with total and average statistics
Fix: Tooltip positioning — Chart tooltips now follow mouse cursor and display above pointer for better readability
Fix: SVG viewBox height — Increased chart height to prevent Y-axis labels from being cut off

2.30.0

New: Schema Settings UI — Added 4 admin controls for schema settings that were previously registered but had no UI: Auto-Output Schema, Validate on Save, Default Schema Type, AI Schema Provider
New: Default Schema Type select — Choose which schema type to use by default when auto-generating (Article, FAQPage, HowTo, etc.)
New: AI Schema Provider select — Choose which AI provider to use for AI-mode schema generation (OpenAI, Claude, Gemini)
New: Post Type Mapping "Default" option — Post Type Mapping now includes a "Default (Global Setting)" option that falls through to the global Default Schema Type
Fix: FAQPage fallback — When FAQPage schema is generated but no FAQ content is found in the post, it now falls back to Article instead of generating an empty invalid schema
Fix: Schema meta box default type — Post editor schema dropdown now shows the configured Default Schema Type instead of hardcoded "Article"
Fix: Bulk generate schema type — Bulk generation now uses configured defaults instead of hardcoded types

2.29.2

Fix: Auto-clear provider errors on success — Provider errors are now automatically cleared from the Debug log, General tab alert, and AI Providers tab alert when a provider successfully responds. Prevents stale error notifications from persisting after the issue has been resolved.

2.29.1

Fix: Provider card badge — Provider card header badge now uses the current selected provider instead of stale last_working_provider transient. Claude and other providers now correctly show "Active" badge when selected.

2.29.0

New: Provider Debug Log — Dedicated Debug tab in Settings shows all AI provider errors with timestamps, provider names, and error details. Stores last 50 errors in wp_options with auto-pruning. Clear All button to reset the log.
New: Provider Error Alerts — Red error alert appears below the selected AI provider on General and AI Providers tabs when errors are detected. Shows the latest error message with a direct link to the Debug tab.
New: Provider Status Redesign — Provider Status badges now show 4 clear states: Active (green), Error (red with inline error text), Configured (gray), Inactive/No key (muted gray). Removed dependency on stale last_working_provider transient.
Fix: OpenAI & Claude error handling — Both providers now return structured error arrays (['error' => true, 'message' => '...', 'status' => int]) matching Gemini/OpenRouter pattern, ensuring Failover_Handler catches and logs errors correctly.
Fix: Debug details in analytics only — Provider error details (e.g., HTTP 429, invalid API key) are appended only to the analytics log database, not sent to the frontend chatbot. Chatbot shows a clean user-facing message.
Fix: Double-encoded HTML entities — Fixed esc_html__() in Failover_Handler causing I'm to display as I'm in chatbot responses. Changed to __() with escaping at render time.
Fix: Analytics Show More button — Fixed mismatched CSS class selectors (.sentientai-show-more → .sai-show-more) and added analytics page to JS enqueue condition so the expand/collapse functionality works.
Fix: Provider errors persisted across saves — Removed sentient_ai_provider_errors from Settings API registration to prevent WordPress from resetting the error log to empty on every form save from other tabs.

2.28.0

New: Conversation Export (CSV) — Export all conversations or a single conversation as CSV from the Analytics > Conversations page
New: Conversation Export (PDF) — Export a single conversation as printable HTML (Ctrl+P to save as PDF) with styled chat bubbles and timestamps

2.27.1

New: Schema JSON-LD Copy Action — Added Copy JSON-LD buttons to the post editor schema meta box and dedicated Schema editor for faster schema testing workflows
New: Google Rich Results Test Link — Added one-click Rich Results Test links for posts with schema markup
Fix: Schema Editor Action Layout — Schema editor action buttons now wrap cleanly inside the editor card on narrower admin screens

2.27.0

New: Custom AI System Prompt — Admin-configurable system prompt textarea in Features tab to override the default AI behavior, tone, and response rules. Empathy Engine modifiers still apply on top
Fix: Privacy Mode chatbot error — Removed invalid typingIndicator reference in privacy mode block that caused ReferenceError and prevented chatbot from responding
Fix: Corrupted fuse.min.js — Replaced truncated Fuse.js v7.0.0 (3.9KB) with complete minified build (23.8KB) to fix Local Brain fuzzy search

2.26.0

New: Smart Auto-Reply Suggestions — AI generates 2-3 context-aware follow-up question chips after every bot response. Suggestions use sentiment, topic, query keywords, and reply content for relevance
New: Context-Aware Suggestion Engine — generate_suggested_questions() method in API_Handler uses sentiment-based, topic-based, WooCommerce-aware, and keyword-based patterns to produce relevant follow-ups
Update: Chatbot Frontend — appendMessage() now renders clickable suggestion chips below bot messages. Clicking a chip populates the input and sends immediately
Update: API Response Format — All 11 response paths now include suggested_questions key in JSON payload
Update: Follow-Up Chip Styling — Chips use filled primary color hover with white text and subtle shadow for clear interactivity. Long text wraps at word boundaries within 220px max-width

2.25.0

New: Token & Cost Tracking Dashboard — SVG usage chart with 7/30/90 day period selector for daily token consumption and estimated cost
New: Budget Alerts — Configurable spending limits with visual progress bar on dashboard. Set daily/weekly/monthly thresholds
New: Model Breakdown Table — Per-provider, per-model token/cost/requests breakdown on dashboard
New: Cost-per-Query Stat — Average cost per AI request shown on dashboard
New: CSV Export — One-click export of usage data with provider, model, tokens, cost, latency columns
New: Enhanced Usage Stats — Daily activity table now shows tokens, cost, and average latency columns
Update: Logger::get_usage_by_model() — New method for model-level usage analytics

2.20.1

Fix: Token tracking pipeline — Provider clients (OpenAI, Claude, Gemini, OpenRouter, HuggingFace) now return usage data from API responses; Failover_Handler accepts success arrays; API_Handler uses correct token keys
Fix: GEO Scanner schema detection — Auto-generated schemas from post publish now use consistent _schema_type/_schema_data meta keys, making them visible to GEO scoring
Fix: GEO Scanner FAQ quick fix — Generates JSON-LD schema instead of microdata; stores _schema_data for proper scoring; uses hidden <details> element instead of visible placeholder text
Fix: GEO Scanner HowTo detection — HowTo quick fix now only offered when content contains step-by-step patterns (numbered headings, "Step 1", sequential language)
Fix: GEO Scanner i18n — Word count now uses Unicode-aware regex for CJK, Arabic, Hindi, and other non-Latin scripts
Fix: GEO Scanner security — Scan action now requires nonce verification to prevent unauthorized database writes
Fix: Database migration — Logs table columns added automatically on upgrade (previously required table recreation)

2.22.0

New: SEO/GEO/AEO Positioning — Plugin rebranded as "WordPress SEO Domination: AI-Powered GEO, AEO, Schema & Semantic Search". Chatbot is now positioned as a bonus feature, not the main focus
New: SEO-Optimized Product Responses — HTML-formatted product responses with semantic markup (schema.org attributes) for search engine indexing
New: JSON-LD Schema for Product Search — Automatic ItemList + Product schema generation for multi-product search results and comparisons
New: AEO-Optimized Formatting — Featured snippet-ready response format with direct answers, ordered lists, and comparison tables
New: GEO Scanner Product Analysis — Extended GEO scanner to analyze WooCommerce products with 10 product-specific factors (description, images, attributes, categories, tags, reviews, price, SKU, FAQ)
New: Schema Injection — Chatbot responses automatically inject JSON-LD schema into page head for search engine indexing
New: Product Response HTML Rendering — Chatbot frontend renders semantic HTML for product responses instead of plain text
Update: WooCommerce_Integration — Added format_product_response_html(), format_multi_product_response_html(), format_comparison_response_html(), generate_product_list_schema(), generate_comparison_schema() methods
Update: API_Handler — Product responses now include reply_html and schema_json fields for SEO optimization
Update: chatbot.js — Added injectSchemaJsonLd() function and updated appendMessage() to render HTML for product responses
Update: chatbot.css — Added styles for product cards, comparison tables, and AEO-optimized layouts

2.23.0

New: FAQ Gutenberg Block — Native Gutenberg block for creating FAQ sections with questions and answers. Supports accordion, tabs, and simple list layouts with search functionality
New: FAQ Schema Auto-Generation — FAQ block automatically outputs FAQPage JSON-LD schema in wp_head for rich results
New: Graph Relation Output — Connected @graph schema linking Organization, WebSite, and content schemas with @id references
New: Organization Schema Settings — Admin settings for Organization name, logo, URL, type, and social profiles
New: WebSite Schema with SearchAction — Automatic WebSite schema with sitelinks search box for Google
New: Enhanced Rich Results Validator — Added recommended properties array with required/recommended badges and completeness scoring
New: Three-tier Site-wide Audit — Schema dashboard now shows Valid/Warning/None classification with completeness percentages
New: FAQ Block Editor Styles — Editor-specific styles for the FAQ block with visual feedback
New: FAQ Block Frontend Styles — Responsive frontend styles with accordion, tabs, and list layouts
Update: Schema_Generator — Added build_schema_graph(), get_organization_schema(), get_website_schema(), output_site_graph() methods
Update: FAQ_Manager — Added register_faq_block(), enqueue_block_assets(), render_faq_block(), render_faq_item_block(), output_faq_schema() methods
Update: validate_schema() — Now returns property-level detail with status, message, and level (required/recommended)
Update: get_schema_status() — Now includes valid, warning, and none counts for three-tier classification

2.24.1

Fix: GEO Scanner form submission — Fixed malformed input, missing nonce, and score display issues
Fix: Autopilot intent classification — classify_intent now correctly matches direct action names (list_posts, check_updates, etc.)
Fix: check_updates command — Fixed missing includes and function name compatibility for WP 7.0.2
Fix: list_posts command — Now shows up to 100 posts with total count
Fix: Schema deletion — Replaced AJAX with proper form POST + nonce + admin_post handler
Fix: Schema SQL queries — INNER JOIN now excludes posts without schema; empty types filtered out
Fix: API key preservation — API keys no longer vanish when saving from non-provider tabs
Fix: Provider status badges — Dynamic green/gray/red badges now reflect actual working provider state
Fix: Provider card active state — Active badge now matches the last working provider (via failover)
Fix: Failover settings layout — Content no longer renders outside the card-body container
Fix: Failover provider display — General tab now shows a dynamic "Failover active" notice when a different provider is handling requests, matching the AI Providers tab status
Fix: Autopilot execution safety — Added try/catch safety net to prevent fatal errors
Fix: Provider tab field styling — Fixed Delay and Welcome Message field styling on the providers page

2.24.0

New: SERP Preview in Post Editor — Google search result card preview directly in the post editor meta box. Shows title, URL, description, and rich result previews for FAQ, Product, HowTo, and Article schemas
New: Live SERP Preview — SERP preview updates in real-time as you edit post title and excerpt
New: Rich Result Previews — Visual previews for FAQ dropdowns, Product cards, HowTo steps, and Article cards
New: Validation Details Panel — Post editor meta box now shows completeness percentage, errors, and warnings inline
New: FAQ Block JavaScript — Rewrote FAQ block JavaScript to use plain WordPress API instead of JSX for better compatibility
Update: Schema_Generator — Enhanced render_schema_meta_box() with SERP preview and validation details
Update: FAQ_Manager — Fixed block registration to use block.json files and proper script enqueuing

2.21.0

New: Enhanced Product Search — Multi-result product search with natural language filter parsing. Find products by color, size, brand, price range, category, and sort order
New: Product Comparison — Compare products side by side with "compare X and Y" or "X vs Y" queries. Shows price, rating, stock, and attribute differences with best value indicators
New: Attribute Filtering — Filter products by any WooCommerce attribute (color, size, brand, material, etc.) using natural language like "red shoes under $50" or "Nike size 10"
New: Store Filters API — New methods to retrieve available categories, attributes, and price ranges for the store
New: Product Suggestions — Get product suggestions based on partial input for autocomplete functionality
New: Comparison Detection — AI detects comparison intent in queries (compare, vs, versus, difference between, which is better)
New: Filtered Search Detection — AI detects filtered search intent (under/over price, color/size/brand mentions, sort keywords)
Update: WooCommerce_Integration — Added search_products(), compare_products(), suggest_products(), get_store_filters(), detect_comparison_query(), is_filtered_search(), parse_search_filters() methods
Update: API_Handler — Product query handling now checks for comparison and filtered search before single product lookup

2.20.0

New: AI Provider Failover — Automatic fallback chain when primary AI provider fails. Configure failover order, skip providers without API keys, and log failover events
New: Token & Cost Tracking — Database logging of token usage, provider, model, latency, and estimated cost per AI request. Dashboard displays monthly totals and per-provider breakdowns
New: Scheduled Log Cleanup — WP-Cron job to automatically purge old log entries. Configurable retention period (7-365 days), optional file log clearing, manual cleanup button
New: Settings Import/Export — Export all plugin settings as JSON and import from file. Supports full configuration transfer between sites
New: Dashboard Token Stats — Three new dashboard cards showing monthly tokens used, estimated cost, and average latency
New: Provider Usage Breakdown — Dashboard card showing per-provider token usage, cost, and request count
New: Provider Status Indicators — Failover settings section shows API key status for all providers with color-coded badges
New: Auto-Generate Schema — Automatically generate schema when posts are published for the first time. Configurable per post type
New: Schema Post Type Mapping — Set default schema type for each WordPress post type (Post→Article, Page→WebPage, Product→Product)
New: Enhanced WooCommerce Schema — Product schema now includes aggregateRating, brand, SKU, and offer data when available
Update: API_Handler — Refactored to use Failover_Handler for provider routing with automatic retry
Update: Logger — Extended database schema with tokens_used, provider, model, latency_ms, and estimated_cost columns

2.19.0

New: Schema Markup Generator — AI-powered JSON-LD schema generation for posts and pages. Supports 16 schema types (Article, FAQPage, HowTo, Product, LocalBusiness, Event, Organization, Person, VideoObject, BreadcrumbList, and more)
New: Three generation modes — AI mode (uses OpenAI/Claude/Gemini), Auto mode (rule-based, no API key needed), and Manual mode (edit JSON-LD directly)
New: Schema validation — Real-time validation of JSON-LD markup against schema.org requirements with required property checks
New: Post Editor meta box — Schema status, type selector, and one-click AI/Auto generate buttons directly in the post editor sidebar
New: SERP Preview — Google search result preview in the schema editor showing title, URL, description, and rich result previews for FAQ and Product schemas
New: Dashboard stats — Schema Markup overview (posts with/without schema, type breakdown) and GEO Scanner overview (average score, last scan) cards on the main Sentient AI dashboard
New: Schema Generator admin page — Dedicated Tools sub-tab with schema dashboard, per-post editor, bulk generation, and type filtering
New: Admin chat commands — Autopilot Agent now supports "generate schema", "schema status", and "validate schema" commands
New: Frontend JSON-LD output — Automatic <script type="application/ld+json"> output in wp_head for posts with schema data
New: Bulk schema generation — Generate schema for multiple posts at once with type selection and auto-detect mode
Update: GEO Scanner — Enhanced schema detection now validates JSON-LD data integrity and offers more quick fixes (FAQPage, HowTo, regenerate via AI)
Update: GEO Scanner — Fixed schema detection ordering (FAQ check runs before schema factor to enable FAQ-specific quick fixes)

2.18.0

Fix: AI provider error handling — HuggingFace, Claude, and OpenAI clients now return real API error messages instead of generic "unable to process" response
New: HuggingFace router endpoint — Switched to OpenAI-compatible router endpoint (router.huggingface.co/v1/chat/completions) for broader model support
Update: Model selections — Added latest models: DeepSeek-R1, Claude Sonnet 4.5, GPT-4.1 series to provider dropdowns
Fix: Dashboard card duplication — Removed duplicate System Status and Quick Actions cards on the assistant settings page

2.17.0

New: API error notifications — Structured error responses from all AI providers (Gemini, OpenAI, OpenRouter, HuggingFace, Claude) shown as admin notices with provider, model, error message, and timestamp
New: Email alerts on API failure — First API error per incident sends email to admin; auto-clears on successful retry
New: Custom privacy notice — Configurable privacy notice text in Privacy tab; displayed in chatbot or falls back to default
New: Model dropdowns — OpenRouter and HuggingFace providers now have model selection dropdowns in settings
New: Gemini header auth — Updated Gemini Client to use x-goog-api-key header authentication for broader compatibility
Fix: Proactive Chat field duplication — Fields no longer duplicate per provider in settings UI
Fix: Form validation on delay input — Delay field no longer blocks save
Fix: Provider reset on save — Provider selection no longer resets when saving Providers tab
Admin menu restructured — All submenu items (Contact Manager, FAQ Manager, GEO Scanner, MCP Server, Setup Wizard, Usage Stats, Vector DB) consolidated under Sentient AI parent with updated redirects
FAQ Manager menu nesting — FAQ menu now properly nested under Sentient AI parent
Session management enhanced — Conversation handler with improved session lifecycle, webhook validation hardened in WooCommerce integration
Vector DB — Stability improvements and query optimization

2.14.0

CRITICAL: Server-side session secrets — Replaced client-side SHA-256 session hash with server-generated session secrets stored in wp_options, preventing session forgery and cross-site session hijacking
CRITICAL: SQL injection fix in Vector DB — Admin dashboard query stats now uses escaped table names, closing SQL injection vector in the vector database status panel
HIGH: WooCommerce webhook hardening — Webhook secret is now required (returns 403 if unset), missing signature returns 401, and unknown webhook event types are rejected with 400
HIGH: Session lifecycle management — Added session registration, verification, and destruction methods with automatic pruning of sessions older than 24 hours
Fix: Chatbot clear history — Session secret is now properly reset and re-initialized after clearing conversation history

2.13.0

CRITICAL: Atomic rate limiting — Replaced transient read-then-write with atomic $wpdb->query() increments, eliminating race condition that allowed unlimited requests under concurrent traffic
CRITICAL: Session hash verification — Conversation history and clear endpoints now require SHA-256(session_id + home_url) proof, preventing cross-site session spoofing
HIGH: Conversation history sanitization — History messages are now sanitized through Prompt_Sanitizer before AI calls, closing prompt injection vector via stored conversation data

2.12.0

Performance: GEO Scanner — Capped post dropdown to 200 posts to prevent memory exhaustion on large sites
Performance: Vector DB search — Reduced local search from 500 to 200 rows to lower memory usage per query
Performance: Logger file writes — Switched to fopen append mode eliminating read-modify-write race condition
Performance: Topic Clusterer — Limited topic comparison to top 50 topics, reducing O(n^2) overhead
Performance: Chatbot history loading — Batch DOM append in loadDbHistory eliminates 50+ individual reflows
Security: Session IDs — Replaced Math.random() with crypto.getRandomValues() for cryptographic randomness
Security: Clear history error handling — Added AJAX error handler and secure session ID regeneration
Fix: Deprecated jQuery — Replaced deprecated e.which with e.key for keyboard events
Fix: Admin UI — Added debounce to canned responses input to prevent excessive serialization
Fix: WordPress.org compliance — Reduced readme.txt tags from 27 to 5 (recommended maximum)

2.11.0

CRITICAL: MCP Server authentication hardened — Server now rejects all requests when no API key is configured (fail-closed). Removed insecure body-parameter API key transmission. CORS restricted to site origin.
CRITICAL: MCP sensitive tools require admin capabilities — Contacts and WordPress management tools now require manage_options capability.
CRITICAL: Fixed broken SQL in Usage Stats — Dashboard widget and stats page no longer produce invalid SQL when no date filter is applied.
CRITICAL: Removed unauthenticated learning stats endpoint — nopriv AJAX handler for learning statistics removed.
HIGH: Scoped cache clearing to plugin transients only — Autopilot clear_cache no longer deletes transients from other plugins.
HIGH: Fixed OpenAI API timeout handling — Added missing timeout on message creation step, added WP_Error check on run response.
HIGH: SQL injection pattern fixed — SHOW TABLES queries now use transient-based caching instead of direct interpolation.
HIGH: Conditional file loading — Admin-only files no longer loaded on frontend pages, reducing memory footprint.
HIGH: Query length enforcement — User queries capped at 2000 characters to prevent token exhaustion attacks.
HIGH: Prompt injection logging — Blocked injection attempts are now recorded in the database for audit.
HIGH: FAQ endpoint rate limiting — Added per-IP rate limiting and reduced max FAQ count for unauthenticated requests.

2.10.1

WooCommerce Deep Integration — HPOS, product vectorization, real-time webhooks
HPOS (High-Performance Order Storage) support — reads orders from custom wc_orders tables when enabled, auto-falls back to legacy wp_posts
get_order_stats_hpos() queries either HPOS tables or legacy storage based on what's active
hpos_enabled() detection method for plugin-wide awareness
Product vectorization — auto-generates rich embeddings from product title, description, attributes, categories, tags, reviews, FAQs, variations, and upsells
on_product_save() hooks into woocommerce_new_product and woocommerce_update_product
reindex_all_products() bulk reindex with woocommerce_product_import_end integration
Real-time WooCommerce webhooks — REST endpoint at /wp-json/sentientai-woo/v1/webhook with HMAC signature verification
8 webhook events: order created, completed, cancelled; product updated; stock status changed; review created; low stock; no stock
Event data includes structured payloads with formatted messages for logging
Product webhooks trigger automatic re-vectorization
Secret key configuration with optional signature validation
All events logged to query log with source='webhook'
New WooCommerce Deep Integration card in Features tab with 4 toggles and HPOS/vectorization info
Fixed: $wpdb global used before declaration in handle_webhook_event()
Fixed: Missing register_setting() for sentient_ai_woo_webhooks_enabled and sentient_ai_woo_webhook_secret
Added: sentient_ai_woo_webhooks_enabled to checkbox preservation array
Added: 72 test cases covering all WooCommerce integration methods
Conversational Commerce Engine — AI shopping assistant with Vibe Search, Visual Shop, Gift Concierge
Vibe Search: natural language product matching via Vector DB with tokenized match reasons
Visual Shop: image URL → AI description → visual product matching with domain security
Gift Concierge: 5 guided questions → personalized gift guides with budget/personality scoring
Admin toggles for each feature + allowed image domains list
Fixed: Vibe search used get_post() instead of wc_get_product() for vector results
Fixed: Visual Shop didn't check WooCommerce is active before loading products
Fixed: Match reasons checked full query against product name instead of individual tokens
Fixed: Gift Concierge questions missing translation wrappers
Fixed: Extra PHP bracket in Commerce Engine feature card array

2.8.0

New: Vector Database with two-tier semantic search — Local MySQL (shared hosting) and Pinecone/Supabase cloud engines
New: Auto-generated embeddings for posts, pages, and FAQs with OpenAI API or built-in hash fallback
New: Semantic search integrated into chatbot responses for meaning-based content matching
New: Admin settings page with engine selector, cloud provider configuration, and index status dashboard
Combined 2.7.0 features: MCP Server (8 tools, SSE transport, API key auth, CORS headers)
Combined 2.7.0 features: GEO Scanner (10-factor content analysis, scoring dashboard, one-click schema fixes)
Combined 2.7.0 features: Browser-based FAQ matcher with keyword and title matching
Fixed: Settings reset when saving across tabs (Enable Chatbot, Pre-Chat Form, role limits)
Fixed: Pre-Chat Form data not persisting across page loads (sessionStorage)
Fixed: Welcome message not displaying when chat opens
Fixed: Role limit field blocking form submission when feature is disabled
Fixed: SSE endpoint infinite loop (timeout added)
Fixed: Missing content format in MCP manage_wp tool
Fixed: Unhelpful error message when no API key configured
Fixed: FAQ matching skipped when conversation history exists
Chore: Added Help tab with complete feature documentation

2.7.0

MCP Server — Model Context Protocol support for AI agent integration
8 tools: chat, FAQ search/list/create, contacts, stats, low confidence, WordPress management
3 resources: FAQ database, usage stats, contacts
2 prompts: site overview, FAQ improvement
API key authentication, SSE transport, CORS, rate limiting
GEO Scanner — 10-factor content analysis for AI search visibility
Factors: content length, structured data, FAQs, citations, statistics, E-E-A-T, structure, meta, lists, titles
Scoring dashboard with auto-fix actions
Browser-Based FAQ Matcher — Local FAQ search without API key
Bug fixes — Multi-tab settings persistence, pre-chat form stability, welcome message display
Help tab — Complete feature reference with documentation

2.6.0

Security Hardening — Remediated Assistant ID leak, added output sanitization
Design Refresh — Modernized UI with CSS variables, improved SVG rendering
Bug Fixes — Resolved ReferenceError, history persistence, transient counting, approval flow

2.4.0

WooCommerce Deep Integration — Product lookup by name/SKU, order status, recommendations, cart info
Performance: FAQ query caching, Options_Cache helper

2.3.0

Widget Branding & Theming — 6 themes, custom colors, border radius, live preview
Prompt Injection Protection — 20+ injection pattern detection

2.2.0

Local Brain — Fuse.js-powered fuzzy FAQ search in the browser, zero API costs, offline support
Privacy Mode — Restrict all processing to local database

2.1.0

WordPress Autopilot — 10 commands: updates, broken images, posts, security, cache, stats, health

2.0.0

Sentient Memory — Topic clustering, feedback tracking, auto-FAQ generation, learning dashboard

1.4.0

Empathy Engine — Emotion detection and dynamic tone adaptation

1.3.0

Glass Box Transparency — Confidence scoring and source attribution

1.2.0

Multi-turn conversation context and server-side history

1.1.1

Security and performance hotfixes

1.1.0

Session-based chat history, TXT export, configurable position

1.0.0

Initial release with multi-AI support

Sentient AI — WordPress SEO Domination: AI-Powered GEO, AEO, Schema & Semantic Search

标签

下载

详情介绍:

安装:

屏幕截图:

升级注意事项:

常见问题:

更新日志: