birdeye

BirdEye - Privacy-First AI Assistant

BirdEye is a cross-platform desktop AI assistant that works entirely offline by default, supporting code analysis, multiple-choice question solving, and general Q&A through screenshots and voice input.

Features

Core Capabilities

• Coder Mode: Code explanation, debugging, and snippet generation
• MCQ Mode: Multiple-choice question solving with reasoning
• Question Mode: General Q&A and text summarization
• Overlay UI: Always-on-top, resizable, transparent window
• Voice Input: Local speech-to-text processing
• Screenshot Analysis: OCR-powered content extraction

Privacy & Security

• Local-first processing - All data stays on your device by default
• Encrypted storage - AES-256 encryption for local data
• Explicit consent - Permission requests for all sensitive actions
• GDPR/CCPA/HIPAA compliant - Privacy by design
• No stealth features - Fully transparent operation

Accessibility

• WCAG 2.1 AA compliant - Full keyboard navigation
• Screen reader support - Proper ARIA labels and focus management
• High contrast mode - Supports system accessibility preferences
• Configurable shortcuts - Customizable keyboard shortcuts

Tech Stack

• Desktop Framework: Electron (Node.js + Chromium)
• Frontend: React + Tailwind CSS
• Local LLM: Ollama/llama.cpp (GGUF quantized models)
• OCR: Tesseract.js
• STT/TTS: Web Speech API + platform voices
• Storage: Encrypted local storage with SQLite

Installation

Prerequisites

• Node.js 18+
• npm or yarn

Development Setup

# Clone the repository
git clone https://github.com/your-repo/bird-eye
cd bird-eye

# Install dependencies
npm install

# Start development server
npm run dev

Building for Production

# Build the application
npm run build

# This creates installers in the dist/ directory:
# - Windows: BirdEye Setup.exe
# - macOS: BirdEye.dmg  
# - Linux: BirdEye.AppImage

Usage

Global Shortcuts

• Ctrl+Shift+S - Capture screenshot for analysis
• Ctrl+Shift+V - Start/stop voice input
• Ctrl+Shift+O - Toggle overlay mode
• Ctrl+Shift+B - Focus BirdEye window

Modes

Coder Mode

Perfect for developers who need quick code reviews or explanations:

• Screenshot code snippets from any IDE or browser
• Get instant analysis, bug fixes, or documentation
• Voice queries like "What does this function do?"

MCQ Mode

Ideal for students taking online quizzes:

• Capture multiple-choice questions from slides or PDFs
• Get step-by-step reasoning for the correct answer
• Transparent operation - no cheating features

Question Mode

General-purpose Q&A and summarization:

• Screenshot text passages for summarization
• Voice queries about on-screen content
• General conversational AI assistance

Overlay Window

• Drag to reposition anywhere on screen
• Resize to fit your workflow
• Click-through mode for presentations
• Minimize to system tray

Privacy & Compliance

BirdEye is built with privacy-first principles:

• Local Processing: All OCR, voice recognition, and AI inference happens on-device
• Explicit Consent: Permission dialogs before accessing camera or microphone
• Data Minimization: Only necessary data is collected and processed
• Encryption: All stored data is encrypted with AES-256
• Audit Logs: Complete transparency of all data access
• Right to Deletion: Easy data export and complete removal options

Compliance Features

• GDPR Article 25 (Privacy by Design)
• CCPA consumer rights support
• HIPAA-ready mode for healthcare environments
• Export compliance for encryption software

Configuration

AI Models

BirdEye supports multiple AI providers:

Local Models (via Ollama):

• Llama 3 8B (recommended for general use)
• Mistral 7B (faster, good for code)
• CodeLlama (specialized for programming)

Cloud Models (opt-in only):

• OpenAI GPT-4o
• Google Gemini
• Azure OpenAI
• Anthropic Claude

Model Storage

Local models are stored in:

• Windows: %APPDATA%\BirdEye\models\
• macOS: ~/Library/Application Support/BirdEye/models/
• Linux: ~/.config/BirdEye/models/

Development

Project Structure

src/
├── main/           # Electron main process
│   ├── main.js     # Application entry point
│   └── preload.js  # Secure IPC bridge
├── renderer/       # React frontend
│   ├── components/ # UI components
│   ├── contexts/   # React contexts
│   └── main.jsx    # React entry point
└── shared/         # Shared utilities

Key Components

• MainInterface - Primary desktop interface
• OverlayInterface - Floating overlay window
• ChatInterface - Message display and input
• BirdEyeContext - Global state management
• PrivacyContext - Privacy controls and audit logging

Building Components

All components follow accessibility best practices:

• Semantic HTML structure
• ARIA labels and roles
• Keyboard navigation support
• Focus management
• Screen reader compatibility

Roadmap

Phase 1 - MVP ✅

• Overlay UI (Electron + React)
• Local OCR + LLM integration
• Screenshot capture functionality
• Coder Mode basic flow
• Local encryption and privacy controls

Phase 2 - Expansion (In Progress)

• MCQ & Question Mode implementation
• Local STT + TTS integration
• Multi-model selector with smart routing
• Click-through overlay mode
• Advanced settings UI

Phase 3 - Polish

• Advanced model management (download/remove)
• Custom prompt templates
• Cross-platform installers
• Performance optimizations
• Plugin system

Contributing

• Fork the repository
• Create a feature branch (git checkout -b feature/amazing-feature)
• Commit your changes (git commit -m 'Add amazing feature')
• Push to the branch (git push origin feature/amazing-feature)
• Open a Pull Request

Development Guidelines

• Follow React best practices
• Maintain accessibility standards (WCAG 2.1 AA)
• Add tests for new features
• Update documentation
• Respect privacy-first principles

License

MIT License - see LICENSE file for details.

Acknowledgments

• Ollama - Local LLM serving
• Tesseract.js - OCR processing
• Electron - Cross-platform desktop framework
• React - UI framework
• Tailwind CSS - Styling framework

Support

• 📧 Email: [e-mail]
• 🐛 Issues: GitHub Issues
• 📖 Documentation: Wiki
• 💬 Discussions: GitHub Discussions

Note: BirdEye is designed for productivity and learning. It includes no stealth or anti-detection features and should not be used to circumvent academic integrity policies or monitoring systems.