Context engineering means giving an AI the useful information it needs to do a particular job. That can include instructions, relevant files, examples, and the results of earlier steps.
Imagine asking someone to fix a broken shelf. They need to know which shelf, what is wrong, and which tools they may use. An AI also does better when the task and important background are clear.
More information is not always better: unrelated or outdated details can distract the AI. The aim is to provide the right information for this task, not to send every file or message you have.