Task, Context, Output: the only three parts a prompt has

Every good prompt says what to do, what to work with, and what shape to answer in. See a weak email prompt and a better one side by side, then run the better one.

In this lesson you will

Name the three parts of a prompt
Tell a weak prompt from a better one and say why
Run a three-part prompt and check the result

People talk about prompting as if it were a skill with a vocabulary. It is mostly one habit. When you ask a person for help you tell them what you need, give them what they need to do it, and say what you want back. AI is the same, except it will not ask you the obvious question when you leave a part out. It will guess.

Figure 1.Task, Context, Output. Leave one out and the model fills it in with the most average choice it can find, which is usually not yours.

The same request, weak and better. The better one is longer only because it says the things the weak one left the model to guess.
Part	Weak prompt	Better prompt
Task	"Write an email about the meeting"	"Write an email asking my team to confirm attendance at Friday's planning meeting"
Context	(none)	"It is at 2pm in the small meeting room. Three people missed the last one because the invite went to spam. I want them to reply by Wednesday."
Output	(none)	"Friendly, under 90 words, with the date and time in the first line."

The weak prompt gets you a generic email about "the meeting" with placeholder brackets to fill in. The better one gets you something you can send. Nothing clever happened. You told it what you knew.

Try it

Run the better prompt

Three free runs a day on the site key. Add your own key for unlimited runs.

Knowledge check

Which part is missing from this prompt: "Summarise this article. Here is the text: [article]"?

Lesson FAQ

What makes a good AI prompt?

It says what to do (task), gives the model what it needs to work with (context), and describes the shape of the answer (output). Most weak prompts are missing the second or third part.

Do I need to learn prompt engineering?

Not for everyday use. The Task, Context, Output habit covers most of what advanced techniques achieve for normal tasks. Specialist methods matter for building software on top of AI, not for asking it to write an email.

Does the order of the parts matter?

Slightly. Task first tells the model what kind of answer is coming; context next gives it the material; output last is the final instruction it reads. It is also the order that is easiest for you to remember.

Lesson FAQ

Stop overpaying for AI tools