HKUDS/LightRAG ? reverse-engineered prompt

Build me a complete Retrieval-Augmented Generation (RAG) system. The goal is to create an application that can ingest a large number of documents in various formats like PDFs, images, and text files, and then let users ask questions about that content.

I want it to be smart, so it should build a knowledge graph from the documents to understand the relationships between different concepts, not just do a simple keyword search. When the system generates an answer, it must include citations so we can see which source document the information came from.

Please create a simple web UI for this. Users should be able to use it to ask questions and get answers, and it would be great if there was some way to visualize the data or the knowledge graph. The whole thing should be built in Python and be easy to set up and deploy using Docker. I'll need to connect it to my own database, so make it compatible with common ones like PostgreSQL or even a graph database like Neo4j.