This doesn't seem obviously true.Because what is knowable is the same for everyone.
I don't know the place where you live: that can't tell me that it's not "knowable" for you. If you have access to better information about your home town than I do, this does not point to a different kind of knowing, just a different set of data on your side.
Agnosticism can be rational if it remains a modest, personal claim of not knowing; but it becomes irrational if it claims to know what others can know without knowing what they have indeed experienced or the data to which they may have access that the agnostic does not.
Which you would say are what? I'm guessing one is Descartes "cogito." What's the second?in the sense of knowing beyond even unreasonable doubt, there is an entire thread dedicated to the two irrefutable facts that philosophers have found in two and a half millenia.
Such knowledge is not empirically available anywhere. It is only available in maths, or else in analytic statements. And in both cases, this is because, as you say, knowledge is a kind of definitional or tautological circle, referring back to an abstraction not reality. Knowledge of reality itself is always inductive, not deductive. So to suppose that a Theist would be irrational to believe in something in the absence of deductive certitude would be irrational.knowledge in the sense that it cannot be seriously doubted.
Again, the most the Agnostic position can say is, "I don't know." But it cannot with any conviction or justice say, "You don't know," and certainly not, "It cannot be known by anyone." It simply cannot know these latter two things.