Rather than designing a genie to exactly match your moral criteria, the simple solution would be to cheat and use yourself as the genie. What the Outcome Pump should solve for is your own future satisfaction. To that end, you would omit all functionality other than the "regret button", and make the latter default-on, with activation by anything other than a satisfied-you vanishingly improbable. Say, with a lengthy password.
Of course, you could still end up in a universe where your brain has been spontaneously re-wired to hate your mother. However, I think that such an event is far less likely than a proper rescue.
This assumes that the laws governing the magical universe must be parsimonious, like those of our own. There is no reason why this has to be the case, though. Picture writing a simulation of such a place. Include as many absurd special cases as you want. You could model it on a high level - from the point of view of conscious agents only, and delve into particle interactions only when there is no other choice (say, the inhabitants learn to build particle accelerators.) Even in the latter case, you could stop the sim-creatures from delving too deeply via "slow zone" effects: if they try to build anything complex enough to stress your simulation's resources or reveal embarrassing inconsistencies, it will simply fail to work.