Select2LLM

Select2LLM lets you send selected text to Ollama with a shortcut and paste the result back, making translations, corrections, summaries and code tasks fast and simple.

Features

• Send selected text with a shortcut to your preferred Ollama model.
• Multiple shortcuts and prompts configured by the user to cover different workflows.
• Simple in‑app settings (language, temperature, UI zoom, etc.).
• Model list shown in alphabetical order.
• Thinking filter for reasoning models (shows only the final answer).
• Light/Dark themes and multilingual UI (English, Spanish, Catalan).
• Quick cancel from tray icon or overlay button.

System Requirements

• Linux:
  • For X11, it requires xclip and xdotool.
  • For Wayland, it requires wtype and ydotool.
• Windows: PowerShell must be installed and enabled.
• macOS: Support is planned for future versions.

Installation

1. Dependencies

• Ollama with at least one model:

  ollama pull llama3.2

• Linux packages:
  • X11: sudo apt-get install xclip xdotool
  • Wayland (e.g. Fedora): sudo dnf install wtype ydotool
• Windows: Ensure PowerShell is installed and enabled.

2. Installing Select2LLM

1. Clone

   git clone https://github.com/Len4m/select2llm.git

2. Install

   cd select2llm && npm install

3. Run / Build

   npm start
   # or
   npm run build

Configuration

Shortcut and Prompt Configuration

Manage everything from the app. Advanced users can edit ~/.select2llm/shortcuts.json:

[
  {
    "ctrl": true,
    "shift": true,
    "alt": true,
    "key": "t",
    "prompt": "Translate the following text to English. Return only the English translation without any additional comments or explanations:",
    "model": "llama3.2:latest",
    "temperature": 0.7,
    "overlay": true
  }
]

Global Configuration

Global settings in ~/.select2llm/config.json:

{
  "language": "en",
  "temperature": 0.8,
  "keep-alive": 5,
  "host": "http://127.0.0.1:11434",
  "uiZoom": 100
}

How to Use

1. Select any text on your computer.
2. Use a configured keyboard shortcut to send the selected text to the pre-configured LLM in Ollama.
3. The processed response will be automatically pasted.

Important

• Do not interact with or move focus away from the destination application while Select2LLM is typing; losing focus can cause unexpected behavior.
• You can cancel the current streaming/typing process by double-clicking the tray icon or using the cancel button in the overlay window.

Recent Technical Improvements

• Smart Thinking Filtering: Automatically detects and filters content between <think> and </think> tags from models with reasoning capabilities.
• Service-Oriented Architecture: The application has been refactored with a modular and robust architecture, improving communication with Ollama, configuration management, and error handling.
• Model Organization: Models are now displayed alphabetically for easier navigation.
• Accessibility and Usability: Includes an interface zoom slider for better readability, with changes applied in real-time.
• UI/UX improvements: Polished configuration interface with light and dark themes.
• Multilingual interface: Available in English, Spanish and Catalan.
• Quick cancellation: Double-click the tray icon or use the overlay cancel button to stop streaming.

Contributing

Select2LLM is my first application in Electron, so there are likely bugs and areas for improvement. All contributions are welcome!

Future Enhancements

• Cross-platform support: Improve Windows support and complete macOS.
• Output options: Choose between simulated typing or clipboard paste.
• Automatic updates: Built-in auto-update system.

License

This project is licensed under the GPL-3 License. See the LICENSE file for more details.