5.1. Crafting Effective Prompts
💡 First Principle: A good prompt does the model's hardest job for it — it removes ambiguity about intent and supplies the context needed to act. Everything else about prompt crafting is a tactic for achieving those two things.
Why this matters: the exam asks you to recognize well-structured prompts, explain how Copilot determines context, distinguish zero-shot from few-shot, and apply best practices. These are practical, scenario-friendly topics — you'll be shown a weak prompt and asked how to improve it.
The mental model: giving directions to a capable stranger. They can drive, but they can't read your mind. Tell them the destination (intent), hand them the map of where you are (context), and they'll get you there. Mumble "go that way" and you'll end up somewhere plausible but wrong.
⚠️ Common Misconception: "A longer, more detailed prompt is always better." Clarity, specificity, and relevant context matter far more than length. Irrelevant detail dilutes the context and can make suggestions worse, not better.
