New Features LLM Integration - AI-powered OCR: Use GPT-4o, Claude, Ollama, or Apple Intelligence for vision-based text recognition - Smart post-processing: Automatically format, correct, or transform captured text with LLM - Separate provider configuration: Choose different providers for OCR vs post-processing - On-device processing: Apple Intelligence runs locally with no API calls (macOS 15.1+) - Network-aware fallback: Automatically falls back to built-in OCR when offline Multi-Region Capture - Batch selection: Select multiple screen regions in a single capture session - Combined results: OCR results from all regions are merged together - Optional LLM processing: Post-process combined text with AI - Up to 50 regions: Capture complex layouts in one go Watch Mode - Continuous monitoring: Monitor a screen region and automatically capture when content changes - Interactive region selection: Draw regions with a visual overlay - Floating control panel: Start, pause, and reselect regions on the fly - Multiple output modes: Append to clipboard, write to file, or stream notifications - Smart detection: SHA256-based change detection with configurable polling interval - URL scheme support: Trigger watch mode via `trex://watch` Capture History - Persistent history: All OCR captures saved with thumbnails - Quick access: Last 5 captures available in menu bar submenu - History window: Browse all captures via `trex://showHistory` - Configurable storage: Set max entries (1–10,000) - One-click copy: Quickly copy any previous capture to clipboard Table Detection - Vision-based detection: Uses macOS 26+ document recognition for table extraction - LLM fallback: Falls back to AI-based table detection on older systems - Multiple formats: Export tables as Markdown, CSV, or plain text - Menu bar toggle: Quick access to table detection settings Bug Fixes CLI Improvements - Fixed CLI not exiting after stdin EOF (#61) - Added back Intel Support - Improved temp file handling with UUID-based paths Language Recognition - Added Turkish (tr-TR) to Vision OCR fallback language sets (#62) - Fixed language code normalization to handle underscore format (#65) UI Fixes - Fixed Quick Setup content overflow with scrollable language grid (#63) - Added confirmation dialog when hiding menu bar icon with recovery instructions (#60)

- Fixed ⌥ symbol not showing in Settings on older versions of macOS - Fixed Notification with captured text goes directly to Notification Center instead of being presented to the user

- New: Add Option-Click on the menubar icon to trigger a custom action - Improvements to faceless mode - Fixed: Settings window presented on TRex launch

- New: Capture from Clipboard menu bar item - Fixed: Fixed missing Settings\Preferences in some cases

- New: macOS Ventura support, including new laguages: Japanes, Korean, Ukranian and Russian - New: Automatic language detection and ability to capture multi-language text. macOS Ventura only. - New: Drag&Drop an image on the menu bar item to OCR it - New: New option in CLI to OCR a file - New: macOS Services integration - New: Integration with Shortcuts. Option to run a Shortcut on successful capture. - New: Dedicated Keyboard shortcuts for capture from screen and capture from clipboard, with or without running automation - New: Custom words supports. If you're using TRex for highly specialized text(medical, engineering, etc.) you can improve text recognition by providing a list of words, which are specific to your field. - New: URL Scheme endpoints for triggering capture from clipboard - New: Embedded CLI tool - New: Official Alfred workflow(download on GitHub) - Fixed: Cleaned up text capitalization in menus and preferences

- New: Support for Spanish, Portuguese and Chinese - Fixed: Crash in diacritical marks processing

- Bug fixes

New: Optional Notification with recognized text. Enable in Preferences.

This version brings macOS Monterey Compatibility and new features. New Shortcuts: - Trigger automation URL - Recognize text from image in the clipboard

Added option to ignore line breaks

You can configure TRex to run actions on text capture: - Auto open URLs found in QR codes - Auto detect and open URLs found in text - Launch a custom URL Scheme, use {text} to pass captured text as parameter. Fixes: - Fixed a typo in URL scheme trex://showPreferences - Fixed a typo in Capture Text menu item