In software testing, particularly blackbox test. How do you deal with cases with unknown expected result?

I was given a task to do a blackbox testing on a command line app. Say the app is supposed to take an integer within certain range and give some output.

I wasn't given any other information apart from this. So then when input is float or an out-of-range integer, the input is not valid but the app could also return something.

So the question I have is: how should I handle those cases. I probably shouldn't assume it should return an exception or something because it wasn't clear when I received the features of what this app is supposed to do. My test (I use Pytest in Python) also can't do assertion as I don't know the expected result.

I would like to keep a record file just with a list of all unspecified cases and their raw outputs for the developer to review. Is that the right way to go?

Many thanks