[XTR] AI Assistant 1.0.35

This release introduces essential bug fixes and improvements covering vector knowledge base responses, the bot selection menu, bot reply reliability when mentioned, and AI image generation model handling.

An issue where vector knowledge base (Vector Store / RAG) responses occasionally included unwanted citation tokens, file reference tags, and source brackets has been resolved. Bot answers are now automatically sanitized, delivering clean and polished responses without leaking internal reference codes.

An issue where clicking the "Ask AI" button in the editor failed to load the bot list when multiple bots were assigned to a forum has been resolved. The bot selection dropdown now opens seamlessly.

Additionally, a permission conflict that previously prevented mentioned bots from replying in certain forums and logging their activity has been fixed. Bots will now reliably answer user mentions in all designated forums.

For image generation, model options in the admin control panel have been clearly organized by provider, and automatic fallback safeguards have been introduced to ensure smooth and uninterrupted image creation.

Changelog

• Automatically stripped file citation tags, source reference brackets, and internal citation tokens from vector knowledge base (RAG) responses.
• Fixed an issue where clicking the "Ask AI" button threw an error when multiple bots were assigned to the same forum.
• Resolved a forum permission conflict that prevented bots from replying when mentioned and missing entries in the activity log.
• Added safeguards to prevent background notification processes from interrupting bot reply workflows.
• Grouped image generation models by provider in the admin control panel for clearer selection.
• Implemented an automatic fallback mechanism to prevent errors caused by mismatched image generation model names.
• Added an upgrade routine that rebuilds the bot cache and normalizes image generation settings upon update.

With this update, we are introducing an intelligent external file search capability that provides significant efficiency and cost savings, particularly for forums that manage extensive documentation, rulebooks, or lengthy manuals.

Previously, when large documents were assigned to a bot, the entire text was loaded and processed on every single reply. For communities with comprehensive knowledge bases, this could lead to excessive token consumption and high operational costs. With version 1.0.34, OpenAI's native vector storage and file search technology is now fully integrated into the add-on.

Instead of reading through entire documents on every prompt, the assistant now searches the stored files in the background and retrieves only the exact paragraphs relevant to the user's question. This keeps conversations accurate, fast, and remarkably lean, slashing token usage by over ninety-nine percent on large documentation lookups.

All existing configurations remain completely unaffected. Bots that rely on the local knowledge base or alternative AI providers will continue operating smoothly without requiring any adjustments.

Change Log

• OpenAI Vector Store Support: Added configuration fields for OpenAI Vector Store ID and maximum retrieved results under the Knowledge Base tab in the Bot management screen.
• Targeted Document Retrieval: Prevented massive documentation from bloating system prompts. Responses now utilize OpenAI's file search tool to extract only relevant context chunks per query.
• Massive Token Optimization: Dramatically reduced per-reply token consumption by over ninety-nine percent for bots managing large rulebooks and extensive manuals.
• Cross-Feature Compatibility: Fully adapted the new file search flow across thread mentions, direct messages, profile posts, automated replies, and support ticket answers.
• Backward Compatibility: Guaranteed zero disruption for bots using the local knowledge base or alternative AI providers without a vector store configured.

With this milestone update, our primary focus has been completely revamping and perfecting the AI Discussion Summary system, making it an indispensable, natural, and lightning-fast part of the forum discussion experience.

---

What's New & Highlighted Features​

1. Native & Modern XenForo Interface
The summary box has been rebuilt from the ground up using XenForo's native core styling. Say goodbye to clunky, heavy custom interfaces! The new card design feels modern, sleek, fully responsive, and naturally integrated with your forum theme across both mobile and desktop.

2. Four-Tier Structured Summary Architecture
Discussions are no longer squished into a single plain paragraph. Readers now get a rich, multi-layered breakdown:

• Executive Summary (TL;DR): A quick 1–2 sentence overview for members on the go.
• Key Takeaways: Bulleted core insights highlighting main arguments and citing specific post numbers.
• Discussion Outcome (Conclusion): Clear consensus or current status of the debate.
• Comprehensive Full Summary: A well-balanced, multi-paragraph narrative covering the full depth of the conversation.

3. Natural Human-Like Text-to-Speech (Listen Feature)
Members can now listen to the entire summary with a single click! Our updated phonetic processor ensures community-specific terms like "XenForo" and "XenTR" are pronounced accurately and naturally without awkward speech interruptions or robot-like glitches.

4. Multilingual Experience with Independent Language Memory
The summary dynamically adapts to the browsing language of each visitor. Multilingual forums can now serve summaries in Turkish, English, German, and more without translations overwriting each other in the database.

5. Lightning-Fast & Resilient Generation for Long Threads
Whether a thread spans 2 pages or 50 pages, the new context extraction and token balance algorithms ensure summaries are produced completely, smoothly, and up to 3x faster without truncation or hitting token roadblocks.

---

Community Feedback Built-in:
Interactive "Was this summary helpful?" buttons allow members to vote on summary quality, giving community managers real insights into reader satisfaction.

Change Log​

New Features​

• Native XenForo Interface: Rebuilt the summary template utilizing XenForo's native core classes, eliminating heavy custom styling in favor of a clean, lightweight card layout.
• Structured 4-Tier Summary Architecture: Summaries are now clearly partitioned into TL;DR, Key Takeaways, Conclusion, and Comprehensive Full Summary.
• One-Click Text-to-Speech Player: Added a built-in audio player allowing members to listen to discussion summaries with intuitive Play/Pause controls.
• Multilingual Summary Storage: Summaries are now stored and presented per language code, automatically aligning with the visitor's active forum language preference.
• Summary Feedback Rating System: Integrated "Was this summary helpful?" voting buttons to gather direct community feedback on summary accuracy.
• Collapsible Content Toggle: Members can easily expand or collapse the summary card to conserve vertical screen space.
• Admin Activity Log Integration: All summary generations and refresh operations are now systematically recorded in the Admin Control Panel activity log.

Improvements & Enhancements​

• Human-Like Pronunciation & Speech Flow: Refined speech synthesis cleaner to ensure brand terms like "XenForo" and "XenTR" are spoken accurately and smoothly.
• Generation Speed & Token Economy Optimization: Streamlined reasoning and token allocation for long discussions, slashing processing times from 35s down to 7-8s.
• Smart Post Sampling for Large Discussions: Enhanced context preparation by prioritizing the original post, top-reacted contributions, and author responses within large multi-page threads.
• Content Hash Caching: Summaries remain cached unless new discussion posts are detected, preventing redundant and costly API consumption.
• Automated Fallback Provider Protection: Seamlessly routes generation requests to an alternative secondary provider if the primary AI service encounters an issue.
• Concurrency Mutex & Flood Guard: Added concurrency locking to safeguard against duplicate simultaneous generation requests on the same discussion.

Bug Fixes​

• Resolved Premature Truncation on Long Threads: Fixed an issue where long or heated discussions could result in summaries being cut off mid-sentence.
• Fixed Summary Duplication Glitch: Resolved a parsing issue where the Full Summary section fell back to repeating the TL;DR text.
• Fixed Cross-Language Overwrites: Fixed database unique constraints so multilingual summaries coexist peacefully without overwriting one another.
• Safety Filter Handshake Fix: Addressed content filtering halts on controversial forum topics, ensuring comprehensive analysis completes reliably.

This release elevates your administrative experience with interactive analytics, high-definition charts, and sleek metric cards while refining the core content generation engine for unmatched reliability.

Key Highlights:

• Next-Gen Administrative Dashboard:
  The administrative interface has received a complete overhaul. Enjoy dynamic time-series activity trend charts with multi-period filters (24 Hours, 7 Days, 30 Days, All Time), high-DPI distribution donuts, executive KPI cards, and an intuitive emergency Kill Switch banner.
• Visual Schedules Overview:
  Automated content schedules are now monitored through elegant metric cards featuring a responsive circular SVG capacity ring, today's run counter, queue status, and next scheduled run indicators.
• Panoramic Activity Log Metrics:
  The activity log now features a live success rate circular gauge, aggregate token consumption with cost estimations, and response latency speed benchmarks.
• Optimized Bot Workflow & Generation Engine:
  Thread creation, profile messaging, replies, and translation services have been optimized. The new thread trigger pipeline was redesigned to prevent duplicate entries and eliminate redundant database queries.
• Knowledge Base & Import Improvements:
  Resolved security rule mismatch when importing exported Knowledge Base SQL files. Multi-category bot assignment now utilizes bulk database operations for maximum throughput.
• UI Polish & XenForo 2.3 Standards:
  Missing and misaligned phrases across the admin control panel have been corrected, toggle tooltips aligned with XenForo 2.3 standards, and navigation icons refreshed.

Change Log

• New Feature: Integrated dynamic time-series activity trend charts with customizable time filters (24 Hours, 7 Days, 30 Days, All Time) into the Admin Dashboard.
• New Feature: Added high-DPI distribution donut charts displaying operation status, action types, AI providers, and active users.
• New Feature: Added executive KPI cards summarizing total operations, success rate, token consumption, and average response latency.
• New Feature: Upgraded the Schedules list with circular SVG capacity rings, today's run counter, queue status, and next scheduled run indicator.
• New Feature: Integrated visual success rate gauges, token consumption with estimated cost calculations, and response speed badges into the Activity Log.
• New Feature: Added an intuitive, one-click global system status banner (Kill Switch) to start or pause AI operations effortlessly.
• Improvement: Optimized bot content generation services for threads, profile posts, replies, and translations to maximize system stability.
• Improvement: Restructured the AI bot trigger workflow on new thread creation, preventing duplicate records and eliminating redundant database queries.
• Fix: Resolved a security rule mismatch that prevented exported Knowledge Base SQL files from being imported via the import tool.
• Improvement: Converted multiple Knowledge Base category assignments for bots into bulk database operations, significantly increasing execution speed.
• Fix: Corrected missing or misaligned interface phrases across category editing, log filters, bot test modals, and Knowledge Base management screens.
• Improvement: Aligned toggle button hover tooltips with XenForo 2.3 user experience standards.
• Improvement: Updated the settings icon in the admin navigation menu in accordance with Font Awesome standards.
• Improvement: Fixed metric counter synchronization on the activity log summary banner.

This update focuses on enhancing daily performance, optimizing your API credit usage, and ensuring seamless compatibility with the modern XenForo 2.3 architecture.

Key Highlights:

• Lightning-Fast Thread Summaries: The thread summarization feature now features intelligent smart-caching. As long as no new replies are posted to a thread, the existing summary is served instantly without repeated AI calls. This eliminates waiting times for your members while saving your valuable AI credits.
• Smarter Conversation Suggestions: Fixed the tone and quote alignment issue when requesting AI reply suggestions in private conversations. The assistant now accurately analyzes the context and provides much more relevant, natural drafts.
• Enhanced Guest Rate Limiting: Strengthened cooldown and daily request limits for guests to prevent spam and protect your AI quotas from unnecessary exhaustion.
• Peak Stability with XenForo 2.3 Standards: The add-on's internal architecture has been fully aligned with XenForo 2.3 native conventions. Legacy duplicate files have been cleanly consolidated, making the add-on lighter, faster, and more robust.

We strongly recommend all administrators install this update to enjoy improved performance and quota efficiency.

• Smart Summary Caching: Introduced two-tier caching for the thread summarization tool. Unless new posts are made to the thread, summaries load instantly without calling external AI services.
• On-Demand Refresh: Added support for refreshing thread summaries when a fresh perspective is explicitly desired.
• Guest Protection & Rate Limits: Reinforced session-based cooldowns and daily limits for guests to safeguard API quotas against excessive requests.
• Cron Queue State Persistence: Provider request intervals in the scheduled queue processor are now preserved across executions, preventing API rate-limit overruns.
• Conversation Reply Suggestion: Resolved a parameter mismatch where quote text and selected tone were swapped in private conversations.
• Form Request Verification: Enforced strict POST-only validation for reply suggestion endpoints to adhere to security standards.
• Knowledge Base Import Security: Implemented strict validation for SQL imports in the Admin CP, ensuring only legitimate Knowledge Base entries can be processed.
• Image Bot Response Handling: Converted raw JSON outputs in the image bot controller to native XenForo view responses for complete lifecycle compliance.
• Architectural Consolidation: Cleaned up legacy duplicate files in controllers, services, and repositories to fully comply with XenForo 2.3 standards and eliminate autoloader conflicts.

In this release, xAI's official Grok Imagine image generation architecture has been integrated into our add-on.

What's New in This Release​

• xAI Grok Imagine Integration: Added a new GrokImageService provider operating via xAI's official image generation endpoint (https://api.x.ai/v1/images/generations).
• Grok Imagine Model Support: Added full support for the latest xAI image models: grok-imagine-image-2.0, grok-imagine-image, and grok-imagine-image-quality.
• ImageFactory Update: Image generation support has been enabled for the xAI Grok provider within Bot capabilities.
• Dynamic Model Fetching (Model Fetcher): Added one-click dynamic model fetching support via /v1/models for the xAI Grok provider in the Admin Control Panel.
• ACP Template Enhancements: Added Grok Imagine models to the image model selection dropdown on the bot edit screen.

With this release, we have introduced massive structural improvements to our AI Assistant add-on, specifically focusing on the Image Generation capabilities! You can now seamlessly integrate and experience the newest generation of multimodal models offered via OpenRouter (including the GPT-5 Image series, Google Gemini Image suite, and Meta Muse Spark)!

Furthermore, trigger flexibility has been dramatically improved! We completely eradicated the frustrating issue where bots would accidentally reply with conversational text descriptions instead of generating actual drawings when prompted. Bots will now instantly identify your drawing commands anywhere inside a prompt and deliver stunning artwork!

[NEW] OpenRouter Multimodal Image Engine (message.images Array Parser):
Engineered an advanced parsing module capable of detecting and extracting generated graphics from the modern message.images data arrays returned by state-of-the-art OpenRouter chat/completions endpoints (such as GPT-5 Image and Gemini Image). Seamlessly supports both Base64 Data URIs and remote HTTP image urls!

• [NEW] Added Latest Supported OpenRouter Image Models:
  Revamped the image model selection dropdown in the Admin Control Panel (Bot -> Capabilities tab) to include highly economical and lightning-fast modern vision models:
  • OpenAI: GPT-5 Image Mini, GPT-5 Image, GPT-5.4 Image 2
  • Google: Gemini 2.5 Flash Image, Gemini 3.1 Flash Image, Gemini 3 Pro Image
  • Meta: Muse Spark 1.1
• [ENHANCEMENT] Smart & Flexible Trigger Pattern Detection:
  Removed the strict requirement (str_starts_with) where image triggering keywords (e.g., "create image:", "draw:") had to be positioned at the exact beginning of a sentence. Powered by our new substring scanner (mb_strpos), even if users prepend mentions, greetings, or extra words before the command, the bot instantly intercepts the trigger and routes it to the image creation engine.
• [FIX] Resolved Bots Responding with Text Instead of Drawing Images:
  Fixed an architectural logic flaw where uncaptured trigger prompts would fall through to standard conversational completion APIs, causing bots to describe imaginary photos with plain text instead of drawing them.
• [FIX] Critical ImageFactoryService Bug & Silent Failure Handling:
  Resolved a critical backend bug where generating images via background mentions caused a silent PHP fatal error (Class Not Found) due to an outdated class namespace reference (ImageFactory). Upgraded overall exception handling to capture low-level PHP \Throwable events cleanly without breaking queue execution.

This is a critical maintenance and compatibility update released to comply with the core changes in repository class resolutions (class extensions) introduced in XenForo 2.3.11 and 2.3.12. Additionally, setup errors encountered during new installations have been permanently resolved. It is highly recommended that all users upgrade to this version.

• Fixed: Resolved the TypeError (Return value must be of type...) encountered in Controller and Service files on XenForo 2.3.11/2.3.12 versions.
• Fixed: Fixed the missing AbstractMySql::getData method (Fatal Error) in the Install\Data\MySql file during add-on installation and rebuilds.
• Updated: To ensure full compliance with XenForo 2.3 coding standards, all Repository return types, use imports, and PHPDoc annotations throughout the add-on have been refactored to use the new *Repository suffix (e.g., ProfileRepository).
• Updated: Class calls used for Cache keys have been adjusted to match the new object standards.

This update brings crucial performance and stability improvements to the core architecture of the add-on. The "Log Table Bloat" issue, which could cause problems on large forums, has been completely resolved with the new Two-Stage Pruning system. This ensures your database is free of unnecessary heavy text data while perfectly preserving your bot usage statistics (tokens spent, etc.).
Furthermore, with the newly introduced Auto Model Synchronization, all your AI models (OpenAI, Claude, Gemini, etc.) will always stay up to date effortlessly with the latest frontier versions.

Changelog​

• New: Added Active Days functionality to Schedules. You can now define exact days of the week (e.g. Wednesday and Sunday) for the bot to run, greatly reducing spam and unwanted automated posts.
• New: Added Auto Model Synchronization service.
• New: Added the xentrAiaLogContentRetention option. Heavy text data (prompts and AI responses) is now cleared after a configured period (e.g., 7 days) to save database space, while statistical row data is preserved.
• Enhancement: Completely rewrote the calculateNextRunDate algorithm to intelligently jump to the next selected Active Day, ensuring no wasted background CPU cycles or cron iterations on inactive days.
• Enhancement: Fixed JavaScript logic issues on the "Add/Edit Provider" page in the Admin CP that prevented the API URL and Default Model dropdown from dynamically updating when changing the Provider Type. It now works flawlessly in real-time.
• Enhancement: All provider titles within the Provider.php (Entity) class are now fully integrated into the XenForo Phrase system, ensuring 100% translatability and removing all hardcoded texts.
• Enhancement: Refactored Setup.php by moving all database table creation schemas to a dedicated Install\Data\MySql class to prevent file bloat.

This is a critical maintenance update that resolves significant integration issues with the XenForo 2.3.x core architecture.

In the previous version, attempting to access the add-on's options page in the Admin CP caused a fatal "Class not found (XFCP_Option)" error; this has now been completely resolved. In addition, we have fixed silent failures that prevented front-end AI features—such as AI bot invocation during thread creation, thread summarization, and AI reply suggestions—from triggering. These were resolved by strictly realigning the add-on's class extensions with XenForo 2.3 Controller naming standards.

We highly recommend all users to upgrade to this version to ensure system stability and to restore the full functionality of the AI features.

• Fixed: Resolved the Class "XFCP_Option" not found (Fatal Error) which caused crashes on the add-on options page in the Admin CP. The class extension proxy now correctly inherits XF2 standards.
• Fixed: Resolved mismatch issues in Public Controller class/file naming conventions that prevented AI features (Thread Summarization, Reply Suggestions, etc.) from triggering. (Conversation, Forum, Post, and Thread extensions are now fully aligned with XenForo 2 architecture.)
• Updated: The class_extensions.xml structure has been thoroughly revised to perfectly match the XF 2.3.x core namespace mappings.