Prompting strategy

We mentioned in the previous chapter that you can both use the Chat mode of GitHub Copilot as well as typing inside of a file, and that you’re recommended to use both approaches. As for how to prompt, you’re recommended to write shorter prompts that you add context to as needed. By doing so, you rely on how GitHub Copilot works and how it builds a runtime context based on its underlying model, what’s in your open directory, and your open file, among other things. You will also see another facet of prompting that doesn’t rely on whether the prompt is long or short but rather on how your specific AI tool works. In this chapter, we use the “Exploratory prompt pattern” as described in Chapter 2.