When teams try to improve LLM output, they often start by rewriting the prompt. That can help, but in real systems the prompt is only one part of what the model sees.

Modern LLM applications send more than a single instruction. A request may include a system prompt, retrieved documents, conversation history, tool definitions, user data, and the user's latest question. Together, those inputs fill the context window the model uses to produce an answer.

Context engineering is the practice of choosing and organizing that full input. It means deciding what to include, what to leave out, how to structure the information, how much history to keep, and how to combine retrieved facts with user instructions. In many production systems, better context selection matters more than another round of prompt wording changes.

This chapter focuses on the practical engineering work: token budgets, context types, selection strategies, formatting, conversation history, and the ways long context can fail.