A good friend just released this and it’s very helpful

Fully Local: No external API calls - everything runs on your machine

PDF Support: Recursively processes PDFs from a folder with subdirectories

RAG (Retrieval Augmented Generation): Uses vector similarity search to find relevant context

Streaming Responses: Real-time streaming of LLM responses

Modern UI: Clean web interface built with HTMX and Alpine.js

Single Install: Managed with UV for easy dependency management