You need to install LM Studio for this to work, then open it to network. You will also need to provide your own text corpus. After that, it works more or less like a full-text search engine & summarizer. The summarizer is very flaky, and prone to hallucinations. If that's your thing, then have fun. It will be a wild time.

See my article where I reverse engineer a hallucination here: [TODO]

This uses pretty much only standard library stuff, with the only package being Weaviate, which is the vector database. I am using the vector database only to store embeddings I calculate from LM Studio's models. There is a pre-cooked way to do this in Weaviate. I prefer home cooking, personally, so this system controls the vectorization process itself.