microsoft/graphrag ? reverse-engineered prompt

I want to build a tool that helps me make sense of a large collection of text documents. It should use an LLM to read through all the files and create a knowledge graph of what's inside. The system needs to automatically identify key entities like people, companies, and concepts and then map out the relationships between them.

The end goal is a system where I can ask complex questions about the information scattered across all those documents. The AI shouldn't just do a simple text search; it should use the graph it built to provide deep, summary-style answers and help uncover themes or hidden connections in the data. This will be a Python project. You can use a command line interface for indexing the data and asking questions. Feel free to look up the best libraries for this kind of graph-based RAG system.