What you described is pretty much exactly how modern provable cryptography field works today. Reduction to assumptions about certain primitives and constructing

What you described is pretty much exactly how modern provable cryptography field works today. Reduction to assumptions about certain primitives and constructing the boundaries of security given such assumptions. E.g. "Assuming AES block cipher is a perfect pseudorandom function, prove AES-GCM construction is secure up to certain number of messages"