Tags: anyme123/claude-workbench
Tags
Release v4.0.14 - Auto-update and Dark Theme Fixes Key Features: - Auto-update functionality with signing keys - Dark theme UI improvements - Enhanced contrast and visibility - Fixed extended thinking feature Technical Changes: - Configured Tauri updater with signing keys - Added latest.json generation in GitHub Actions - Fixed hard-coded colors in multiple components - Improved floating element backgrounds - Enhanced text contrast in dark mode Bug Fixes: - Fixed extended thinking parameter name - Fixed dark theme display issues - Fixed auto-update configuration
Release v4.0.13 - Extended Thinking Fix Key Changes: - Fixed extended thinking feature with correct parameter name - Updated to use official --max-thinking-tokens parameter - Improved compatibility with Claude Code CLI standards - Bug fixes and stability improvements
Release v4.0.11: Version update notification system New Features: - Auto-check for updates on startup - Update badge in Topbar - Update dialog with download progress - One-click install and restart - Smart reminder management Technical: - 541 lines of new code - Full TypeScript/React implementation - Tauri plugin integration - 0 errors, 0 warnings Status: Production ready
Release v4.0.10: Critical process management and memory leak fixes - Eliminates all process and memory leaks - Windows Job Objects + Unix process groups - Active child process cleanup - Four-layer cleanup guarantee - 100% process cleanup on shutdown
Release v4.0.9 - Skills Support and Code Block Fixes Critical Fixes: 1. Skills message support (exclude from revert index) 2. User prompt unescape (double-backslash to newline) 3. Code block text selection enabled Skills Message: - Detect and exclude from revert index - Format display with XML parsing - Hide revert button on Skills - Frontend/backend logic aligned Escape Fix: - Replace \\\\n (double backslash) with newline - Correct regex for all escape sequences - Multiline prompts display correctly Code Block Selection: - Add codeTagProps with userSelect: text - Enable manual text selection - Ctrl+Double-click works - Manual copy with Ctrl+C works All issues identified and resolved correctly!
Release v4.0.8 - API URL Normalization Fix Critical Fix: - API URL trailing slash handling - Prevents double-slash in API calls - Affects both OpenAI and Gemini formats Problem: - User enters: https://ai.xiangcao.de/v1/ - Code adds: /chat/completions - Result: https://ai.xiangcao.de/v1//chat/completions (double slash) - API call fails Solution: - Automatically remove trailing slash from apiUrl - Normalize before concatenating path - Works with or without trailing slash Examples: - https://api.com/v1/ -> https://api.com/v1/chat/completions - https://api.com/v1 -> https://api.com/v1/chat/completions Both work correctly now! All API URL formats now handled properly!
Release v4.0.7 - Production Fixes and Enhancements Critical Production Fixes: - Code block copy now works (Tauri clipboard command) - CSP allows all HTTPS (support custom API URLs) - Clipboard write/read commands added - arboard dependency for cross-platform clipboard Features: - write_to_clipboard command (backend) - read_from_clipboard command (backend) - Visual feedback: 已复制! (2s) - Fallback to browser API if needed - Cross-platform clipboard support CSP Update: - Changed from whitelist to allow all HTTPS - Supports custom API URLs (proxies, mirrors) - Official APIs: OpenAI, Gemini, Deepseek, Qwen - Custom APIs: any HTTPS endpoint - Production = development behavior Bug Fixes: - Copy button paste issue resolved - Production API fetch errors fixed - Custom API URLs now work - Long code blocks copy successfully All production issues resolved!
Release v4.0.6 - Third-Party API Prompt Enhancement Major New Feature: - Third-party API support for prompt enhancement - Multi-provider configuration system - OpenAI and Gemini format support - Encrypted API key storage Supported Providers: - OpenAI (GPT-4, GPT-4o, GPT-3.5-turbo) - Google Gemini (gemini-2.0-flash-exp) - Deepseek (deepseek-chat, deepseek-reasoner) - Qwen (qwen-max, qwen-plus, qwen-turbo) - SiliconFlow (Qwen/Qwen2.5-72B) - Any OpenAI-compatible API Features: - Full CRUD for provider configs - Quick preset templates - Test API connection - Enable/disable providers - Optional temperature/maxTokens - Dynamic menu with enabled providers - Settings integration Critical Fixes: - Duplicate project bug (double-backslash paths) - Button nesting HTML error - Delete confirmation dialog - Infinite loop on settings navigation - Event listener fixes All major features fully implemented and tested!
PreviousNext