Smart Hebrew Rtl Ui Safe

NEW IN VERSION 2.2.4
* Claude.ai DOM update: selectors refreshed for Anthropic's new message structure (data-testid="user-message" + .font-claude-response). Hebrew RTL keeps working after the class rename that broke other extensions.

NEW IN VERSION 2.2.3
* Real-time RTL: instant Hebrew detection during streaming -- no more waiting for page refresh
* Flicker-free streaming: eliminates the LTR-to-RTL jump during AI response generation (NotebookLM, etc.)
* Two-phase streaming: CSS-only RTL during generation, full JS processing after completion
* Smarter mixed content detection: English labels followed by Hebrew text now render RTL correctly
* ChatGPT Deep Research RTL: Hebrew content in Deep Research reports now renders correctly -- the only extension that handles this triple-nested sandboxed iframe
* Claude artifact support: Hebrew in TXT, MD, and HTML/JSX artifacts now renders RTL
* Per-site toggle: enable/disable from the popup for any supported site
* Expanded platform support: NotebookLM, Lovable.dev, Base44, Monday.com added
* Native Shortcuts: Right Ctrl + Right Shift to force RTL, Left Ctrl + Left Shift to force LTR
* Security Lock: strict Content Security Policy (CSP) prevents all network access
* Performance: no more page freezes on heavy conversations
* Zero code-block flash: CSS injected before JS runs -- code stays LTR from first paint
* Smart retry: catches Hebrew text that frameworks render with a delay
* Smart input RTL: fixes the unicode-bidi bug where mixed Hebrew/English lines stay LTR in textareas
==============================
FEATURES
* UI-Safe Mode: Unlike other extensions that flip the entire website (breaking sidebars and buttons), this only affects the text you write and read.
* Developer Friendly: Keeps code blocks (```) and syntax highlighting in LTR, even inside Hebrew conversations.
* ChatGPT Deep Research: the only extension that applies RTL inside Deep Research reports.
* Claude Artifact RTL: Hebrew content in TXT, Markdown, and HTML artifacts renders correctly.
* Smart Mixed Content: Lines like "Error 500: השרת נתקל בשגיאה" are detected as Hebrew and displayed RTL.
* Native Keyboard Control: Right Ctrl + Right Shift for RTL, Left Ctrl + Left Shift for LTR. Works like native Windows direction switching -- the only extension that brings this to web platforms.
* Smart Input Fields: Fixes the unicode-bidi:plaintext bug that breaks mixed Hebrew/English lines in textareas. Lines like "Snapshots יומיים" render correctly.
* Privacy First: Built with a "Local-Only" architecture. No analytics, no cloud syncing (besides between your own user), no data collection.
* Platform Support: Optimized specifically for ChatGPT, Claude.ai, Google AI Studio, NotebookLM and more.
==============================
ABOUT
Writing Hebrew on modern AI platforms is often frustrating. The cursor stays on the left, periods jump to the start of the sentence, and mixed English/Hebrew text gets scrambled.
Smart Hebrew RTL fixes this by injecting the correct CSS properties only where needed. It works automatically in the background, or you can take manual control with the new keyboard shortcuts.
Just click "Add to Chrome," refresh your chat, and type naturally.
==============================
PRIVACY & SECURITY
We take privacy seriously.
We have implemented a strict Content Security Policy (CSP) that forbids the extension from connecting to the internet.
Zero Data Collection: We do not track your keystrokes, chats, or usage, nor do we want to.
Zero Transmission: The extension is incapable of sending data to external servers.
Local Processing: All logic happens 100% on your device.
Note: The warning "Read and change data on all websites" is required for the extension to modify the text direction CSS on the pages you visit. It does not mean we are reading your messages.
==============================
SUPPORT
Found a bug? Have a feature request?
Asaf Lecht / אסף לכט
Email: asaflecht@gmail.com
LinkedIn: linkedin.com/in/asaflecht
B'ezrat Hashem, na'aseh v'natzliach