Reverse engineered prompt

Build me a Rust based GraphRAG app that can take one or more text documents, turn them into a knowledge graph, and let me ask normal language questions about what is in the docs. I want it to feel usable right away with sensible defaults, so a simple CLI should work without making me write config first, and it should save its workspace so I can come back and keep querying.

If possible, give me the three ways this project describes, a server mode with an API, a browser only mode that runs fully client side, and a path toward a mixed setup. The browser version should have a basic UI where I can upload a document, see that a graph was built, and ask questions. If local LLM support is available, wire it in as an optional upgrade, but make sure there is a fallback that still works without it.

Please include a small example and clear run instructions, and look up current docs online if you need to.

Want more depth? Deep Reverse