locally infere a random commit message for your staged files and push it immediately. use at your own risk. Models Put your llama.cpp compatible models in ./models. Currently testing with Llama3 and others.