🐛 Bug Fixes

SearchBox

• Fixed search functionality: Resolved an issue where the search box was not working properly
  • Fixed type errors in autocomplete handlers that prevented search from functioning
  • Corrected event handler logic for search input
  • Improved TypeScript type safety to prevent similar issues

Full Changelog: v0.6.0...v0.6.1

🌐 Internationalization (i18n)

• 多语言支持（英文/中文）
• i18next 基础设施
• 语言选择器

AI Capabilities, Export Engine & Build Modernization

🤖 AI Enhancements

• Thinking Mode: Added a thinking mode toggle in the toolbar and popup, enabling models that support reasoning to show their thought process before the final response
• Streaming Preview: Implemented streaming preview functionality with real-time reasoning support and improved UI updates
• Upgraded AI SDK: Bumped the Vercel AI SDK to ai@6, refactoring background streaming to handle reasoning/thinking deltas
• New Providers: Added OpenAI-compatible provider helpers and raw streaming support for providers such as DashScope/Qwen, Zhipu, and MiniMax; updated provider registry with new Qwen provider and refreshed model lists

📤 Export Engine

• Markdown Export: Added markdown export functionality with improved URL cleanup, making it easy to copy article content to note-taking apps
• PDF / PNG / Clipboard Export: Introduced a full export UI (Original vs AI) backed by robust utilities for PDF, PNG, clipboard, and markdown — with cross-origin safety handling, font inlining, adaptive scale calculation, and taint-error retry logic
• CORS-Safe Image Export: Cross-origin images are now fetched via the background service worker and converted to data URLs before html2canvas runs, eliminating canvas-taint failures on pages with external images
• Enhanced Clipboard Copy: Clipboard copy now includes the article title for richer paste output

⚡ Build & Tooling

• Rspack Migration: Replaced webpack with Rspack (@rspack/core ^1.7.8) for significantly faster extension builds; added rspack.config.js and updated dist outputs and vendor chunk naming for both Chrome and Firefox manifests
• Popup Refactor: Lazy-loaded heavy UI components (AIToolbar, RssSubscription, SaveDetailPanel); extracted active-tab messaging helpers; converted callbacks to useCallback for better performance

🐛 Bug Fixes

• OpenAI Stream Parsing: Fixed handling of non-JSON payloads in OpenAI stream extraction to prevent parse errors
• XHR Response Handling: Fixed an issue where accessing XHR response data with responseType='json' could throw an error in certain cases

🎨 UI Improvements

• Model Selector: Redesigned with a unified gradient border container; replaced the icon for the selected model indicator for a cleaner look
• Export Button: Updated icon to IosShareIcon and tightened menu item padding for a more compact feel

RSS Feed Performance & Stability Improvements

🔧 Server Optimizations

• HTTP Client Caching: Enhanced HttpUtils with connection pooling and client management for RSS feed requests, reducing resource overhead and improving fetch performance
• Memory Configuration: Increased Docker JVM max heap size to 1024m for better stability under load
• Connection Timeout: Ensured connection timeout is properly configured in application.yml

🧩 Extension Improvements

• Dev Mode Indicator: Added DEV flag to context menu titles for easier debugging during development
• Code Cleanup: Simplified ArticlePreview title rendering logic by removing redundant isXTwitterSite function

Save Details & Core Backend Improvements

🧩 Extension Support Enhancements

This release rolls out the necessary backend support to power the new features introduced in the latest browser extension update:

• Save Detail API: Added a new PATCH /api/page/{id}/detail endpoint. This allows the extension's new Save Detail panel to seamlessly update saved page metadata (title, description, and source URL).
• Safe Re-saving Logic: When the extension recaptures a page that already exists in your library, the server now avoids overwriting existing core content fields. This ensures any earlier text edits or previously formatted data are safely preserved.

🔧 Server Operations & Fixes

• Performance Tuning: Increased the default JVM heap limits in the Docker configuration and tuned server executor thread limits to optimize overall application responsiveness under heavy loads.

Save Details & UX Improvements (server update required)

✨ Save Detail Panel & UX Enhancements

A significant update bringing new ways to interact with saved content directly from the extension:

• Save Detail Panel: Edit page metadata (title, description), assign collections, and select specific parsers before saving or re-saving content.
• Save Status Badges: Added badge indicators so you can see at a glance if the current tab's URL is already saved in your library.
• Context Menus: Added right-click context menu entry points for opening Reading Mode directly from the page or selection.

🤖 AI Models & Connection

• Provider Updates: Updated DeepSeek descriptions/models and changed default Gemini ID to gemini-3-flash-preview.
• Connection Testing: Improved provider connection testing by correctly supporting empty-string responses from models.

🌐 Localization

• Multilingual Support: Localized the extension name and description for both English and Chinese.

🔧 Under the Hood improvements

• Readability: Bumped @mozilla/readability to v0.6.0 to improve parsing speed and accuracy.

🐛 improve server connection handling

• Add graceful handling when Huntly server connection fails, showing
  warning and allowing AI features to work with external providers
• Hide Huntly AI option in toolbar when server is not connected

AI-Powered Web Extension & Enhanced Twitter Controls

🤖 AI-Powered Web Extension

A major new feature that brings AI content processing directly to your browser extension:

• AI content processing: Summarize, translate, extract key points, and more with streaming AI support
• Multiple AI providers: Configure your preferred AI provider (OpenAI, Anthropic, etc.) directly in extension settings
• Multilingual support: Built-in system prompts for multiple languages with customizable templates
• Article preview: Real-time preview with AI processing results before saving
• Defuddle integration: Enhanced content extraction using Defuddle parser

🐦 Enhanced Twitter/X Auto-Save

New controls for fine-tuning which tweets get automatically saved:

• Minimum likes filter: Set a threshold to only auto-save tweets with a minimum number of likes
• X settings page: Dedicated settings page in the web client for Twitter/X configuration
• Concurrent save prevention: Prevent duplicate saves when multiple capture requests occur simultaneously
• Server-side optimization: Moved minLikes lookup to server-side for reduced API calls and better performance

🔧 Technical Improvements

• Separate extension releases: Independent versioning and release workflow for the browser extension
• Improved excerpt extraction: Better content parsing from HTML for article summaries
• X/Twitter URL detection: Automatic identification of Twitter URLs for special handling
• Caching optimization: Added caching in GlobalSettingService for improved performance

Full Changelog: v0.5.6...v0.5.7

AI-Powered Web Extension

🤖 AI-Powered Web Extension

A major new feature that brings AI content processing directly to your browser extension:

• AI content processing: Summarize, translate, extract key points, and more with streaming AI support
• Multiple AI providers: Configure your preferred AI provider (OpenAI, Anthropic, etc.) directly in extension settings
• Multilingual support: Built-in system prompts for multiple languages with customizable templates
• Article preview: Real-time preview with AI processing results before saving
• Defuddle integration: Enhanced content extraction using Defuddle parser

Sorting & Data Consistency Improvements

🔄 Simplified Sorting for Unsorted Pages

Improved the sorting mechanism for unsorted (un-collected) pages:

• Simplified sorting: Use createdAt directly instead of complex multi-field fallback logic
• Clearer labels: "Unsorted" pages now display "Created" as the sort criteria
• Better consistency: Removed multi-field sort support for cleaner, more predictable behavior

🛠️ Data Integrity Fixes

• CollectedAt auto-population: Ensure collectedAt timestamp is always set when pages are added to collections
• Capture service fix: Pages captured with a collection assignment now properly set the collected timestamp
• Batch operation reliability: Batch move operations now reliably populate collection timestamps

Full Changelog: v0.5.5...v0.5.6