Skip to content

README.md

Latest commit

 

History

History
67 lines (49 loc) · 1.97 KB

README.md

File metadata and controls

67 lines (49 loc) · 1.97 KB

Kratt

Kratt Screenshot

A lightweight, local-first desktop AI assistant for Linux, powered by Ollama.

Kratt is a simple, draggable desktop widget that provides a chat interface for your local language models. It's designed to be a minimal and convenient way to access AI assistance without relying on cloud services.

Requirements

  • Python 3.10+
  • Fedora Linux (or another Linux distribution with standard command-line tools).
  • Ollama installed and running.
  • The grep and find command-line utilities must be available in your PATH.

Installation & Running

  1. Clone the repository:

    git clone https://github.com/importb/kratt.git
cd kratt

  2. Install the required Python packages:

    python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

  3. ** Install Playwright browsers:**

    playwright install

  4. Pull the required Ollama models: By default, Kratt uses a main model, a vision model, and an embedding model. These can be changed in the settings.

    ollama pull qwen2.5:7b
ollama pull moondream:latest
ollama pull nomic-embed-text

  5. Run the application:

    python -m kratt.main

Configuration

Kratt is configurable through its built-in settings dialog, which can be accessed by clicking the gear icon (⚙️) in the header.

  • Models: The application automatically detects and lists all models available in your local Ollama instance. You can select different models for text and vision tasks.
  • System Prompt: You can edit the system prompt to customize the assistant's personality, behavior, and response format.

The default hotkey is set in kratt/config.py and can be modified there if needed.

License

Distributed under the MIT License. See LICENSE for more information.