I'm writing up lecture notes and attempting to put the language on the most logical footing. I'm hoping to distinguish attacks on the consensus itself from exploits of the protocol, as these seem to be separate categories. For example an empty-block attack is clearly an attack: the purpose is to damage Bitcoin, whereas Finney's attack is not attempting to strike a blow to the consensus, it's an attempt to exploit consensus rules in a way that is more favorable to the attacker.
Am I going against a decade of convention or is there some precedent here?
More generally, is there a definition an "attack"?