billion-token-one-task/Deepgraph ? reverse-engineered prompt

Build me a Python app that can read a bunch of research papers, pull out the important claims, methods, and results, and turn that into a structured evidence graph in a database.

I want it to automatically look for interesting gaps or contradictions in the literature, generate promising research ideas from those patterns, and then track which ideas are worth testing. It should have a loop where new paper data updates the graph, the graph suggests new directions, and experiment results feed back into what gets prioritized next.

Please make it feel like a real research engine, with auditable outputs, stored lineage for every idea, and clear status tracking so I can see what was ingested, what was discovered, and what is still pending. If there’s a simple web UI or dashboard for browsing the graph and the current research queue, that would be great too. Use current docs online if you need to confirm best practices.