X Recommender Plus

Smart post scoring tool that dynamically adjusts content transparency based on LLM scores

## What is X‑Recommender? Tired of your X/Twitter or Weibo feed being flooded with low-value content? X Recommender Plus uses AI to score posts, fade or hide weak content, translate posts inline, and explain the current post in a side panel with optional media understanding. The current version supports both X and Weibo, works with local or cloud models, and now includes multimodal side-panel analysis plus temporary model switching for retries. --- ## Why You Need This Tired of your X feed being flooded with: - Filter out gossip, spam, rage bait, self-promotion, and clickbait automatically - Focus on higher-signal content across both X/Twitter and Weibo - Understand unfamiliar posts faster with built-in translation and side-panel explanations - Use local or cloud models depending on your privacy and speed preferences From now on, your social feed can stay readable, explainable, and under your control. --- ## Key Features ✔ **AI-Powered Scoring** - Uses large language models (LLM) to analyze posts on X/Twitter and Weibo - Scores range from 0-100 based on information value, originality, clarity, and engagement potential - Works with popular LLM provider - New in v2.1.0: optionally prefer Chrome built-in Gemini (Prompt API) for filtering scores ✔ **Inline Translation from Score Menu** - Click the score badge to open a quick action menu near each tweet - Translate both tweet text and quoted tweet text while keeping the original content visible - Default translation targets: English and Simplified Chinese ✔ **Side Panel Explanations with Follow-Up Chat** - Open the current post in the extension side panel and get a concise explanation - Keep asking follow-up questions about the same post in an ongoing chat - Side panel responses can stream progressively for a faster feel - Side panel explanations can now include image and video-cover context from both X and Weibo posts ✔ **Smart Filtering and Reputation Tools** - Low-quality posts automatically become transparent or fully hidden - High-quality content remains fully visible - Keep per-post filter toggle, score detail, and reputation actions in the score menu - Whitelisted accounts can bypass filtering immediately and use platform-scoped IDs for X and Weibo ✔ **Fully Customizable** - Create custom scoring prompts tailored to your preferences - Fine-tune score threshold, hide/opacity behavior, cache duration, translation targets, and engine strategy - Detect whether Chrome built-in Gemini (Prompt API) and Chrome Translator are available before enabling them - Switch between multiple configured models and temporarily override the model for specific side-panel or translation retries ✔ **Privacy-Focused** - All processing happens locally or through your configured API - No data collected by us - Scores and state are cached locally only - Local model setups are fully supported if you want an offline-first workflow --- ## How It Works 1. Install the extension 2. Configure your preferred LLM provider (local or cloud) 3. Browse X as usual 4. Tweets are analyzed automatically in the background 5. Low-quality tweets fade out while valuable content remains clear 6. Hover over a tweet to view its AI score and evaluation --- ## Important Notes - This extension analyzes publicly available user-generated content on X. - X and Twitter are trademarks of X Corp. - This extension is not affiliated with, endorsed by, or sponsored by X Corp. ## 1.5 Updates ### New Features - **ID Whitelist**: Add trusted users to a whitelist - their tweets will not be filtered and display as ∞ - **Filter Mode**: Choose between opacity (fade) or hide completely - **No-Text Tweet Scoring**: Option to score tweets with only images/videos - **Feedback Link**: Added feedback and follow link in the footer - **Three-Tab Settings**: Reorganized settings into API, Filter, and Whitelist tabs ### Improvements - Adjusted retweet badge position ## Version 1.5.1 ### Fixes - Fixed hide mode filtering to only use the hide threshold, preventing overlap with opacity behavior - Improved tweet text fallback extraction (`innerText || textContent`) to reduce unstable hide/show flicker ## Version 1.6.0 ### New Features - Added support for multiple models - you can now configure and switch between multiple LLM APIs - Automatically disable thinking mode with model that support thinking ### Fixes - Fixed issue where adding a second model would not display in the list (race condition and storage sync delay) ## Version 1.6.1 ### Improvements - Pre-hide unscored tweets: New tweets stay hidden until scored, then are shown or hidden by score (reduces timeline jumps) - Changed batch size from 5 to 1 to lower timeline jump probability - Changed temperature from 0.1 to 0 for more stable scoring - Added cache check before scoring to avoid duplicate requests - Added viewport detection to prevent large-scale node removal outside viewport causing scroll anchor jumps - Optimized tweet ID extraction logic ## Version 1.6.2 ### Improvements - Thread pass-through optimization: tweets in the same conversation now reuse the first scored result, avoiding repeated scoring for follow-up tweets - Improved conversation detection by prioritizing `conversation_id` grouping (with safe fallback logic), covering mixed-author thread cases like repost + comment chains - Reduced scroll flicker by disabling opacity transition on cached re-apply paths during timeline refresh/scroll - Moved adaptive viewport timeout controls under hide mode only; these options are now shown only when filter mode is set to hide ## Version 1.6.3 ### Fix - Fix issue where preset APIs like Ollama couldn't be added after changing URL to LAN IP (preset values overriding user input) ### Improvements - Preset API name and URL now prioritize user input ## Version 1.6.4 ### Fixes - Improved LM Studio compatibility for Qwen-style models by making the no-thinking fallback path more robust - Prevented score parsing from accidentally reading reasoning step numbers as final scores ### Improvements - Added an LM Studio setup hint recommending non-thinking models first - Added guidance for advanced users to disable thinking via LM Studio Prompt Template settings ## Version 1.7.0 ### New Features - Added "Clear Cache" button to manually clear all cached scores ## Version 2.0.0 ### New Features - Added a score-badge action menu: clicking the top-right score now opens a menu with quick translation targets (English and Simplified Chinese by default) - Added tweet translation for both main tweet text and quoted tweet text, while preserving original content and rendering translation blocks below - Added translation settings in popup: target languages are now multi-select and persisted ### Improvements - Added translation engine strategy options: Ask on first use, Force Chrome, or Force LLM - Added Chrome Translator API detection and status display in settings, with engine options enabled/disabled by availability - Changed single-click score behavior from direct toggle to menu entry, while keeping "Toggle Filter" and preserving double-click score details ### Fixes - Improved repeated translation behavior by reusing existing translation blocks for the same language, preventing duplicate stacking - Added complete i18n keys for translation UI and status/error messaging in both languages ## Version 2.1.0 ### New Features - Added Chrome built-in Gemini (Prompt API) support check in Filter settings - Added a new toggle to prefer Chrome built-in Gemini for tweet filtering scores ### Improvements - Added automatic fallback to the configured external LLM when Chrome Gemini is unavailable or scoring fails - Kept existing filter behavior unchanged when the Gemini toggle is off - Added full i18n text coverage for the new Gemini filter settings and status labels ## Version 2.2.0 ### New Features - Added reputation-based score badges that accumulate scoring history per account - Added a reputation summary entry in the score menu so you can view scored count, average score, and current badge status for a user - Added a quick profile shortcut from the score menu and reputation modal to open the user's X profile ### Improvements - Badge colors now automatically follow a user's accumulated average score after enough history is collected - Whitelisted accounts now use a dedicated reputation state so the shield-style badge and menu messaging stay consistent ### Fixes - Fixed whitelist reputation handling so whitelisted accounts are excluded from reputation statistics and no longer show misleading reputation data - Improved compatibility for old cached score data by rebuilding reputation summaries from stored detail records when needed ## Version 2.3.0 ### New Features - Added an extension side panel for explaining the current tweet text from the score menu - Added follow-up chat in the side panel so you can keep asking about the same tweet - Added model response streaming for side panel explanations and LLM translations - Added side panel i18n text and Chrome side panel permission/configuration ### Improvements - Updated the extension display name, settings title, and footer branding to X Recommender Plus - Persisted side panel state so the current tweet and explanation can be restored after reopening ### Fixes - Fixed side panel follow-up request handling so later questions keep the current tweet context - Fixed stale side panel loading states by recovering old in-progress states after timeout - Prevented older translation responses from overwriting newer translation requests --- ## Version 2.4.0 ### New Features - Added Weibo support for filtering, score badges, translation actions, and side panel explanations - Added platform-scoped whitelist entries: X uses handles like `x/owen`, while Weibo uses numeric IDs like `weibo/6882232397` ### Improvements - Extended post extraction, score actions, and side panel context so the same core workflow now works across both X and Weibo --- ## Version 2.5.0 ### New Features - Added media-aware side panel explanations for both X and Weibo posts, including image and video-cover previews from main posts and quoted posts - Added multimodal support for local model backends, including local OpenAI-compatible servers, LM Studio, and Ollama, so side panel explanations can include attached media - Added temporary model switching for side panel reruns and translation retries, so one request can use a different configured model without changing the default API ### Improvements - Filtering now defaults to text-only scoring for a faster timeline experience, while media-aware scoring remains available as an opt-in setting - Improved side panel tab isolation, loading states, and media preview behavior so explanations stay scoped to the current tab and current post - Weibo media previews are now hydrated through the page context, improving preview reliability for protected image URLs ### Fixes - Fixed Weibo detail-page scoring after SPA navigation and improved score badge placement on Weibo detail pages - Fixed translation placement for reposts and Weibo detail pages, and made empty or no-op translation responses less likely to be accepted as valid results - Fixed whitelist handling so trusted tweets can apply the whitelist result immediately and stale whitelist-only cached scores no longer persist incorrectly --- ## Version 2.6.0 ### New Features - Added batch model setup: fetch the models available from a configured service and add multiple selections at once - Added a cleanup action for removing model configurations that are no longer available from their service ### Improvements - Improved multi-model management with separate manual and batch-add flows, plus clearer highlighting for the default model - When the same model name is available from multiple services, its source is now shown so you can choose the right model for translations and side panel explanations - Settings now save automatically and take effect immediately, with no separate save action required ### Fixes - Improved compatibility after extension updates, reducing unnecessary error notices when an already-open page has an expired extension connection ---