The entity restricting itself to the issue nearly everyone agrees on, is a middle of the road choice to it, and it'd still be allowing evil from the point of the view of the entity to happen, while also suffering with some of the more extreme definitions of free will and auto determination.
Some of the more extreme definitions of free will and auto-determination are utter bunk (though pretty, and they can be fun to poke at.)
They break down, aren't internally consistent, etc., so forth. We can safely discard those as actually viable, as moral theory more or less has for the last good spat of time, heh.
That said, yes, some interventions would be pretty much considered universally good. These are not the interventions that's actually the issue here. The biggest problem is that on the issues not everyone agrees on.
Mmhmm... in those cases, it's debatable whether you're actually dealing with an
ethical issue or if you're dealing with a
cultural one (and whether you separate the two -- I'd easily argue that you must, because while there are baseline moral actions, what extends beyond that depends on geographic complications, if nothing else.). Largely depends on what exactly the intervention is, in other words. As I've been noting, depends on what, exactly, the entity considers evil
There's fairly solid arguments though, iirc, that extending beyond your basic harm prevention package is indeed immoral if the recipients do not wish it. That's th'ol' "White Man's Burden" thing, really.