I don't see your post--have you removed it?
I would also be hesitant to supervise an undergraduate doing research on people with full NPD.
One approach you could take is to study narcissism in "normal" populations, viewing it as a dimensional construct and taking an individual differences approach. You are sure to find some variance in narcissism in a community sample, using a measure like the SNAP-2, which has a narcissism scale. Perhaps you could approach it this way, making it clear that you are not looking for those who meet diagnostic threshold for NPD?