A model answers from what is in its context window on that call and nothing else. Context engineering is the work of choosing that content: the instructions, the conversation so far, retrieved documents, tool results and memory, all within a token limit.
More context is not always better. Long prompts cost more and run slower, and models tend to use information at the start and end of a long input better than material in the middle, an effect known as lost in the middle. The guides here cover managing the context window in AI agents, where to place documents and questions in long prompts, compressing prompts with tools such as LLMLingua, grounding answers in supplied text, and the agent harness that assembles context on each turn.
The library lists all guides with filters for topic, level, category and reading time.