AI Assistant

This project is a local AI assistant designed to run offline on Windows 11. It utilizes lightweight machine learning models for natural language processing.

Now supports both text and voice input/output, advanced markdown/code rendering, dark mode, and full session/prompt management in the GUI.

Features

• Loads a lightweight LLM (e.g., Phi-3 or Mistral) using HuggingFace Transformers
• Supports GPU acceleration via DirectML or ROCm (if available)
• Accepts text input/output in both terminal and GUI (Tkinter)
• Modular code: core logic in assistant/core.py, configuration in config/settings.py, entry point in main.py
• Model and generation settings are configurable
• Conversation history with adjustable context window (/history N)
• Session management: save, load, and auto-save conversations
• Export conversation as plain text
• Multi-line input support (/long and /end)
• System prompt modes: chill, formal, coder (switch with /mode)
• Prompt preview (/prompt)
• Token/character count display after each response
• Error handling for model and file operations
• GUI: mode switching, history management, and command support
• Command handler: /calc for arithmetic, /read for file snippets, /snippet for code blocks
• Voice input (speech-to-text) and output (text-to-speech) in the GUI
• Advanced markdown/code rendering in the GUI (syntax highlighting for code blocks, bold, italic, tables, etc.)
• Dark mode toggle in the GUI
• Full session management (save/load/export/clear) in the GUI

---

Project Structure

ai-assistant
├── data/            # For markdown files or session logs
├── snippets/        # For Kotlin or code snippets
├── src
│   ├── main.py          # Main entry point of the application
│   ├── assistant
│   │   ├── __init__.py  # Package marker for assistant module
│   │   └── core.py      # Core logic for the AI assistant
│   ├── config
│   │   └── settings.py  # Configuration settings for the application
│   └── utils
│       ├── helpers.py   # Utility functions for various tasks
│       └── commands.py  # Command handler for /calc, /read, /snippet, etc.
├── requirements.txt      # List of dependencies
├── .gitignore            # Files and directories to ignore in version control
└── README.md             # Documentation for the project

Installation

1. Clone the repository:

   git clone <repository-url>
   cd ai-assistant
   

2. Create a virtual environment (optional but recommended):

   python -m venv venv
   

3. Activate the virtual environment:

   venv\Scripts\activate
   

4. Install the required dependencies:

   pip install -r requirements.txt
   

Usage

To run the AI assistant in the terminal:

python src/main.py

To run the graphical interface:

python src/gui.py

GUI Features

• Modern Tkinter interface with ttk themes (light/dark mode toggle)
• Logo/icon support
• Scrollable, color-coded conversation window
• Mode selector (chill, formal, coder)
• Prompt preview/export buttons
• Session management buttons: Save, Load, Export, Clear
• Voice input (🎤) and output (🔊) buttons
• Markdown/code highlighting for assistant responses
• Token/character count display after each response

Key Commands (Terminal & GUI)

• /help → Show available commands
• /calc <expr> → Evaluate basic math (e.g., /calc 12 * (4 + 1))
• /read <file> → Read a markdown or text file from current or data/
• /snippet <file>→ Insert a Kotlin or code snippet from snippets/
• /mode [name] → Switch assistant mode (chill, formal, coder)
• /clear → Wipe conversation history
• /save [name] → Archive current session
• /load [name] → Restore a saved session
• /export [file] → Export conversation as text
• /history [N] → Set number of turns to keep in context
• /prompt → Show the full prompt sent to the model
• /long → Paste multiple lines, finish with /end
• exit, quit → Exit the assistant

---

To use voice features:

• Make sure your microphone is enabled and working.
• For text-to-speech, your system must support pyttsx3 (works on most Windows setups).

To use advanced markdown/code rendering:

• The GUI will automatically highlight code blocks, bold, italic, and tables in assistant responses.

---

Follow the on-screen instructions to interact with the assistant. All features are available in the GUI; the terminal version supports text only.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any improvements or features.

License

This project is licensed under the MIT License. See the LICENSE file for more details.