Changelog

A timeline of every ACLI release — new features, improvements and bug fixes.

v1.1.3 2026-03-02 Latest
New Features
  • Smart Model Router — AI automatically selects the best model for each message based on intent analysis
  • 13 intent categories: code writing, debugging, code review, deep reasoning, tech consulting, creative writing, analysis, knowledge Q&A, chat, quick answer, and 3 video types
  • User avatar now displayed in chat message bubbles (Google account profile picture)
Improvements
  • Status bar streamlined — cleaner Auto indicator without redundant icons
  • Default model set to Auto for new installations
  • Lightweight routing via gemini-2.5-flash-lite with ~200ms overhead and minimal token cost
v1.1.2 2026-03-02 Stable
New Features
  • Sound notification system with 6 built-in sound effects (Default, Bell, Ping, Chime, Pop, Drop)
  • Sound effect selector with live preview in Settings
  • Desktop notification when AI completes a reply (background/minimized only)
  • Update button shake animation in title bar for visibility
  • Model switcher now groups models by Provider for easier navigation
  • Model switcher refresh button to reload online models
  • Browser Agent human takeover banner with Resume AI button
Improvements
  • Notification and sound playback moved to main process — works reliably when app is minimized or in background
  • Sound files cached to temp directory for instant playback on repeat
  • Desktop notifications suppressed when app window is focused (sound still plays)
🐛 Bug Fixes
  • Fixed notification never firing due to React state timing issue
  • Fixed sound not playing (replaced invalid base64 WAV with real MP3)
  • Fixed Browser Agent popup window — migrated from inline WebContentsView to independent BrowserWindow
  • Fixed Resume AI button missing after BrowserPanel removal
v1.1.1 2026-03-01 Stable
New Features
  • Browser Agent — AI-controlled browser with 9 tools (navigate, click, type, scroll, screenshot, etc.)
  • Browser Agent independent popup window (1100×750, positioned next to main window)
  • Human takeover mechanism for login pages and sensitive operations
  • Three-layer browser security: AI prompt rules, sensitive URL detection, manual confirmation
  • Internationalization (i18n) support for 7 languages (EN, ZH-CN, ZH-TW, JA, KO, DE, RU)
Improvements
  • Browser window.open intercepted and loaded in same window to prevent extra popups
  • AI system prompt prevents auto-closing browser unless explicitly requested
🐛 Bug Fixes
  • Fixed browser covering chat bubbles in previous inline layout
  • Fixed second browser open failing after close
  • Fixed viewportHeight=0 and screenshot 0KB issues
v1.1.0 2026-03-01 Stable
New Features
  • Integrated SeeDANCE 1.5 Pro video generation (text-to-video / image-to-video)
  • Audio-enabled video generation (generate_audio toggle)
  • Generated videos are auto-downloaded to the server with inline preview in chat
  • Video player supports playback, open in browser, and right-click save
  • Video params selector above input (duration / resolution / aspect ratio / audio)
  • Reserved SeeDANCE 2.0 model slot for future API availability
  • App version displayed in title bar with automatic update detection
  • Added changelog page
Improvements
  • Ark API Key field added to Admin Volcengine settings
  • Video params embedded via [VIDEO_PARAMS] prefix, hidden from user display
🐛 Bug Fixes
  • Fixed "no such column: today_used" error during token deduction
  • Fixed Ark API endpoint path and request body format
v1.0.0 2026-02-28 Initial Release
New Features
  • ACLI desktop client launch (Electron 33 + React 19 + Tailwind)
  • Multi-model chat support — Claude, GPT, Gemini and more
  • 10 AI tools (file ops, code execution, web search, etc.) via OpenAI Function Calling
  • 20 Sub-Agents + 15 Skills library
  • Jimeng AI Video Generation 3.0 Pro (text-to-video / image-to-video)
  • Google OAuth login + invite code registration
  • SSE streaming chat with Thinking Block display
  • Chat history management and export (Markdown)
  • Custom model configuration (custom API endpoints and keys)
  • File drag-and-drop / image paste / multi-attachment support
  • Syntax highlighting, diff view, file creation cards
  • VIP membership tiers (Free / Plus / Pro)
Architecture
  • Volcengine V4 HMAC-SHA256 signature authentication
  • Zustand state management + LocalStorage persistence
  • macOS DMG installer with Apple code signing
  • ACLI CLI command-line tool (Bun compiled)