But you can reason about what any possible transaction can do by reasoning about all possible inputs to your code; you don't need to know anything about how other contracts were written, since even if they are formally provable, they could send you literally any input.
Edit: Rereading your earlier comment, I understand now - you're talking about other contracts your code calls, not vice-versa. In that case, I'd point out that you're totally free to either write those contracts yourself if they're not already provably secure, or write your own code such that it's formally proved to work regardless of what those contracts do.
