One really easy trick that the author missed that tends to work in my experience: adding some examples before/after the prompt. Eg: # Compute the differenc

One really easy trick that the author missed that tends to work in my experience: adding some examples before/after the prompt. Eg: # Compute the difference between each element in an array. # compute_difference([1, 2, 5, 10]) == [1, 3, 5]) Overall you shouldn't expect Copilot to read your mind. If your description is vague, you're going to get unreliable results. If you have a lot of context, clear descriptions, unit tests, existing code, etc, it has a better chance of guessing your intent.