The best assignment should always be: show me what you can do. The student should approach every assignment as though that were the implicit assignment. Do something good for me.
So the details of the assignment cannot detract from that. A good student will interpret the assignment better and give you something better. A bad student will opt for the less imaginative version of the same thing. A really good assignment would not be capable of an unimaginative interpretation. Some day I will think one up.
I don't remember college professors giving us prompts. You were to write a paper on something related to the course material.
I had another assignment in which the students had to invent an imaginary poet, then compare that poet's poem with the title "Arte poética" to poems of the same title by Borges, Huidobro, Neruda. As you can expect, the best part of these papers was the invention of the apocryphal figure, since the other parts were rehashes of class discussions we had had about those three poems. Students who actually wrote the entire poem, not just enough of it to quote from, also did better. I really tricked them into doing something a bit better than they would have thought to do, something a bit transgressive. We make up all of our analyses anyway, so why does the poem have to be authentic? And all poems are made up anyway, right?
(In the same level class sometimes I'll have them write an extra monologue to a play, or internal monologue to a novel. So you can show that you understand the perspective of a character by doing this, without using the author's own words.)
I realized reading the papers that the students didn't quite get the Arte poética genre, because their invented poems were not metapoetic enough. Students seemed to think that Still, I think it was successful.
And it is plagiarism-proof.
I think my assignments are the best aspect of my teaching.