I hope for local model chat inexperienced users are just using llama.cpp's built-in web server interface, which gives you everything you need. No need for a harness or any other chat client.