instavm/clickclickclick

项目介绍

Currently supporting local models via Ollama (Llama 3.2-vision, qwen3.5:4b), Gemini, GPT 4o. The current code is highly experimental and will evolve in future commits. Please use at your own risk.

The best result currently comes from using Gemini 3.1 Flash-Lite as both planner and finder.

Ollama models: qwen3.5:4b works as planner (slow, basic navigation) but not reliable as finder (poor UI element detection). Use with --image-quality=45 for better performance.

Clone the repository and navigate into the project directory:

It is recommended to create a virtual environment:

Put your model specific settings in config/models.yaml and export the keys specified in the yaml file.

Before…

摘自 github.com/instavm/clickclickclick