Generally, the operator @ need not satisfy the axioms of K (normality), however the obvious version of the problem for normal operators also makes sense.

We know that if @ satisfies K4, then the logic of @ (by the usual fixed-point argument) is closed under the Löb rule:

@A -> A / A
Hence, it contains GL. Thus, for example, S4 is not a logic of any such @.