This seems to be a thin wrapper around libllama, what's the point? Llama.cpp already ships a web server. I don't see anything in the README that llama.cpp doesn't already support.