Every score comes with the model's reasoning on the paper page.
"Maestro: Agentic Formal Proof Development by Coordination" scored 7.5/10 predicted impact on @kurateorg — Clarity 8.0 · Significance 7.5 · Evidence 7.0