J.A.R.V.I.S. - AI Code Assistant

J.A.R.V.I.S. is an intelligent coding assistant that leverages multiple state-of-the-art language models to help you with code generation, modifications, and technical discussions.

Features

• Integrated Terminal:

  • Cross-platform terminal support (Windows and Linux)
  • Automatic workspace directory initialization
  • Full xterm.js integration
  • Real-time output streaming
  • Command history support
  • Proper directory tracking
  • Native shell integration (cmd.exe on Windows, bash on Linux)

• Multi-Model Support: Choose between different AI models for your coding needs:

  • DeepSeek R1
  • DeepSeek V3
  • Codestral
  • Gemini 2.0 Flash Experimental
  • Grok 2
  • Claude 3.5 Sonnet
  • GPT-4 Turbo
  • GPT-4o Mini
  • GPT-4o
  • o1-mini
  • o1-preview

• File Attachment Support:

  • PDF files with text extraction
  • Microsoft Word documents (.docx)
  • Excel spreadsheets with sheet parsing
  • Images with OCR capabilities
  • Enhanced Markdown with GFM support
  • All major programming languages
  • Configuration files
  • Text and documentation files
  • File preview with syntax highlighting
  • Multiple file upload support
  • Progress indicators and file size display
  • Type-specific icons and preview buttons

• Real-Time Updates:

  • WebSocket-based notifications
  • Instant feedback for code changes
  • Real-time workspace updates
  • Automatic change notifications

• Workspace Management:

  • Create and manage multiple workspaces
  • View workspace history
  • Delete workspaces when no longer needed
  • Rename workspaces
  • Browse workspace file structure

• Code Generation & Modification:

  • Generate new code based on natural language prompts
  • Modify existing code with AI assistance
  • Preview changes before applying them
  • View diffs of proposed changes

• Interactive Chat:

  • Discuss code and technical concepts
  • Get explanations about existing code
  • Context-aware responses based on workspace content
  • Attach files for additional context

Technical Stack

• Backend:

  • Flask web framework
  • Flask-SocketIO for WebSocket support
  • Eventlet for async operations

• Frontend:

  • Pure JavaScript
  • TailwindCSS for styling
  • CodeMirror for code editing
  • Socket.IO client for real-time notifications
  • PDF.js for PDF processing
  • Mammoth.js for Word documents
  • XLSX.js for Excel files
  • Tesseract.js for OCR
  • Marked and Unified.js for Markdown

Installation

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt

3. Set up your environment variables in .env:

   DEEPSEEK_API_KEY=your_deepseek_api_key
   CODESTRAL_API_KEY=your_codestral_api_key
   OPENROUTER_API_KEY=your_openrouter_api_key
   GOOGLE_API_KEY=your_google_api_key
   GROK_API_KEY=your_grok_api_key
   ANTHROPIC_API_KEY=your_anthropic_api_key
   OPENAI_API_KEY=your_openai_api_key
   

Usage

1. Start the server:

   python app.py

2. Open your browser and navigate to http://localhost:5000
3. Create a new workspace or select an existing one
4. Choose your preferred AI model
5. Start coding with AI assistance!

Model Capabilities

• DeepSeek R1: Latest DeepSeek model with enhanced code understanding and generation
• DeepSeek V3: Specialized in code generation and modification
• Codestral: High-performance code generation model
• Gemini 2.0 Pro: Advanced code generation and natural language understanding
• Grok 2: Advanced language model for code and natural language
• Claude 3.5 Sonnet: Advanced reasoning and code understanding
• GPT-4 Turbo: Latest GPT-4 model with improved performance and up-to-date knowledge
• GPT-4o Mini: Experimental model with specialized capabilities and unique instruction handling
• GPT-4o: Optimized GPT-4 variant with enhanced code understanding
• o1-mini: Experimental model with specialized capabilities and unique instruction handling
• o1-preview: Experimental model with specialized capabilities and unique instruction handling

Contributing

Contributions are welcome! Please feel free to submit pull requests.

Special Thanks

• Nikole Cardoso for her invaluable contributions and support
• Guilherme Guirro for his expertise and guidance
• Felipe Santos for his dedication and insights

Their contributions have been instrumental in making J.A.R.V.I.S. better.

Platform Compatibility

This application has been tested and confirmed working on:

• Linux (native)
• Windows Subsystem for Linux (WSL 2)
• Windows (native, no admin privileges required)

The application uses directory junctions on Windows to avoid requiring admin privileges, while maintaining symlink functionality on Unix-like systems.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.