Relax. We're not going to explain this chapter like a textbook — just like a friend talking you through it, in plain, easy-to-follow terms.
Let's think about it this way for a second
A prompt is an instruction. Context is what surrounds that instruction. Telling the waiter "cook the fish" is the prompt. The note next to it — fish, salt, no spice — that's the context. The model doesn't keep your notes memorized in its head; it only works with what it can see this turn. That's why context engineering is really the job of deciding "what goes on the table." Don't worry about token counts yet — just look at what's on the plates.
Let's connect it to everyday life
A student asks, "Can I use my phone?" Without context, the model just guesses. With a short excerpt from the handbook included, it answers from the actual text. With a whole week of chat, a full PDF, and a .env file crammed in, it gets a stomachache. This course isn't going to throw window-size numbers at you upfront. Just remember four moves — add, drop, summarize, stop — and pack your bag with those.
Let's try it out together
Prompt: ဖုန်းသုံးလို့ရလား၊ စာအရဖြေပါ
Context: handbook.md / ဖုန်း အပိုဒ်တို
Not: last week's whole chat + 3 PDFs + .envExplain context as a sheet of notes placed on the table.5-Minute Try
Write 3 questions. For each one, note what notes you'd add and what you'd leave out — and why.
One Quick Heads-Up
Piling everything onto the table doesn't make the cook any better. Only serve the dishes you actually need.