I'm guessing "prompt engineering" was when people were very highly focused on just wording the prompt correctly, including the right keywords and using all kinds of text tricks to get better results from the AI answers.

And "context engineering" is less about wording and more about warming up the model before asking the main question e.g. uploading relevant files, setting up baseline facts, goals, etc. and then asking?

Edit: Remember "jailbreak prompts" ? lol