Skip to content

OCWS AI Runner

The ocws-llm-runner is a native C/GTK3 desktop client designed to interface with a local, open-source Python LLM server. It allows you to chat with local AI models directly from your desktop and leverages integrated OCR to read text from your screen.

Features

  • Native UI -- A glassmorphic GTK3 chat interface matching the OCWS ecosystem.
  • Model Management -- Scans for and allows dynamic load/eject of .gguf language models from ~/Models/ without restarting.
  • OCR Integration -- A built-in OCR button triggers screen capture, extracts text using Tesseract, and pastes it into the chat prompt.
  • Session Continuity -- The Python backend manages chat history context across multi-turn conversations.

Installation and Setup

  1. Install Python Dependencies:
pip install llama-cpp-python
  1. Download Models:

Create a folder at ~/Models and download quantized .gguf files (e.g., Qwen2.5-Coder-3B or LLaMA-3.1-8B).

  1. Run the Application:
ocws-llm-runner

See Also

  • llm-runner.md -- Full documentation with API reference
  • configuration.md -- LLM Runner section