Nice work, I've raised a PR that adds support for local LLMs (Ollama, llama.cpp etc...) and a few other enhancements.