·
1 commit
to main
since this release
Magic Cursor v0.0.1 (Multimodal Text & Image Processing System)
This standalone release introduces a full hybrid OCR & vision routing pipeline utilizing Google's Gemini models, persistent configuration management, Windows startup registry boot, and custom glassmorphic HUD loading animations.
🚀 Key Features in this Release:
- 🧠 Hybrid Multimodal Vision Routing: Automatically classifies and routes image-heavy queries (diagrams, UI mockups, designs, logos) or queries containing visual terms to Gemini's vision pipeline (
gemini-2.5-flash), while retaining high-speed local OCR for code blocks and text copies. - ⚙️ Persistent Settings Panel: Access the API key panel from the context menu to input and securely save your Google Gemini API key locally (
%LOCALAPPDATA%\MagicCursor\config.json). - 🔌 Zero-Dependency Portability: Built with
WindowsAppSDKSelfContainedenabled. This package is fully self-contained; simply extract the ZIP and runMagicCursor.exewithout installing any frameworks or installers. - 🚀 Launch at Windows Startup: Added a toggle switch in settings that integrates with the Windows Registry (
HKCU\Software\Microsoft\Windows\CurrentVersion\Run) to automatically launch Magic Cursor silently in the background on PC boot. - 🟢 Active Processing Indicators: Custom mouse pointer changes to a pulsing neon green state, and a glowing green "● AI Active" LED displays in the menu header while backend API requests are executing.
- 🎨 UI Glassmorphism & Universal Contrast: Refined list markdown rendering (using aligned grid columns instead of wrapping panels) and forced text foreground variables to guarantee contrast on light/dark Windows themes.
📦 Installation
- Download
MagicCursor_v0.0.1_win-x64.zipbelow. - Extract the ZIP archive to any directory.
- Double-click
MagicCursor.exeto run.