TAG Prompt Framework: Task, Action, Goal

Study guide

TAG Prompt Framework: Task, Action, Goal

TAG is a short prompt recipe with two common meanings. The first is Task, Action, Goal. You use it for decisions, options and management questions where the goal must lead the answer. The second is Task, Audience, Guardrails. You use it for quick writing such as emails, notes and summaries. This course teaches both versions, when to use each, and how to write the words out in full so the AI never has to guess which TAG you mean.

The course is for busy, non-technical professionals who paste prompts into ChatGPT, Claude or Gemini. You will learn to write a Task line that names a real decision, an Action line that asks for three options with trade-offs and a failure check, and a Goal line that tells the AI which trade-off to optimise for. You will also learn the writing version, where Audience and Guardrails keep tone, length and promises under control. Every lesson includes a worked example from a real job and a common mistake to avoid.

What is a prompt framework?

A prompt is the request you type into an AI tool such as ChatGPT, Claude or Gemini. A prompt framework is a reusable structure for that request: a short checklist of the parts a good prompt contains, usually named with an acronym so it is easy to remember. Why it helps: the AI fills in whatever you leave out with generic guesses. A framework makes sure you include the parts that change the answer, such as who it is for, what you want to achieve and the format you need.

Frameworks make prompts repeatable: you fill the same parts each time, save the prompt as a template and share it with colleagues. They are not magic words: the value comes from the detail you put in each part. They help most when a request is complex or open to interpretation; a quick, simple question can stay short. There are well over sixty named frameworks, and most of them remix the same ingredients: role, task, context, audience, format, constraints and examples.

In this course you learn TAG, which is a fill-in-the-slots framework: an acronym where each letter stands for one part of your prompt. It is best for: decisions, options and management questions where the goal must lead.

Two TAGs, one idea

TAG most often stands for Task, Action, Goal. Task says what the job is, Action says how to approach it, and Goal says what you are optimising for. A second version means Task, Audience, Guardrails. It is popular with beginners because the reader and the limits are easy to picture. Because the same acronym carries two meanings, always write the words out in full. A prompt that says only TAG leaves the AI guessing which version you mean.

Both versions keep prompts short. Three clear lines usually beat a long briefing, because the important words stay visible. A useful habit is to draft the Task line first, then choose Action and Goal or Audience and Guardrails depending on what the answer must serve. If you need a decision with options and trade-offs, use Task, Action, Goal. If you need a finished piece of text, use Task, Audience, Guardrails.

The framework is deliberately light. It does not ask you to write a page of background. That keeps the prompt short and the answer focused, which suits small, contained decisions you can describe in one line. When a decision is big, you add the background yourself or move to a framework with a dedicated context slot.

Task: name the decision

The Task is the decision or job you want done, not the topic. If your prompt says weekly meetings, that is a subject. If it says decide whether to move our weekly meeting to async updates, that is a task. Every task line needs bounding facts: team size, current problems, and deadline. These three facts anchor the model's advice to your real situation instead of a generic company.

Phrasing matters more than people expect. A decision framed as a question gets options and trade-offs. A decision framed as an order gets a single answer with no case against it. Start the task line with a clear verb such as decide, choose, compare or draft. Verbs signal the kind of output you want and reduce the chance of an essay. Keep one decision per task line. Two decisions in one line split the model's focus and usually produce a shallow answer to both.

The task line is the first slot in the TAG prompt. Everything after it, the action and the goal, depends on this line being sharp. If you cannot say the decision in one line, you are probably holding two decisions and should split them.

Action: ask for options with trade-offs

The Action line tells the AI what to do with the task. For decisions, the strongest action is give three options with trade-offs, because it forces a comparison instead of a single answer. A trade-off states what you gain and what you give up for each option. Written side by side, trade-offs turn the choice into your business judgement rather than the AI's guess.

The pre-mortem is the line that asks how each option could fail and how soon you would notice. It is quick to add and it exposes weak options before you commit budget or people. Ordering matters. Ask for the recommendation last, after the options and the failure check, so it is reasoned from the trade-offs rather than anchoring the whole answer.

Asking what should I do feels faster but hides the reasoning. You cannot see what the AI ignored, so you cannot compare paths or defend the choice to anyone else. A reusable action line is: give three options with trade-offs, how each could fail, then your recommendation. Save it and paste it into any prompt where you want a decision rather than a summary.

Goal: what you optimise for

The goal names the trade-off you care about: cost, speed, quality, risk or team morale. It is the line that tells the AI which way to lean when options pull against each other. Different goals give different best options, and that is the point of saying it. The same three suppliers can produce three different winners depending on whether you optimise for cost or for risk.

If you have two goals, rank them. Say cost first, then speed. The first becomes the tie-breaker and the second becomes the fallback, so the AI does not choose an order you did not intend. A goal is a dial, not a switch. You are not turning cost, speed, quality, risk and morale off. You are saying which one you turn up for this decision.

The most common failure is leaving the goal out. The AI then picks a default, usually balance or quality, which sounds safe but often does not match the trade-off your job actually cares about. State the goal honestly rather than leaving it implied.

Task, Audience, Guardrails

The writing TAG has three parts. Task is what you want written, Audience is who will read it, and Guardrails are the short rules the writing must respect or avoid. Guardrails are short rules, not long policies. Examples include no jargon, no promises about dates, and stay under 150 words. They prevent problems before they appear. Without them, the AI guesses tone, length and commitments.

Use Task, Audience, Guardrails for quick writing such as emails, notes, announcements and summaries. Use Task, Action, Goal when you need decisions with options and trade-offs, not a finished piece of text. Keep guardrails few and specific. Too many rules can make the writing stiff or slow to produce.

A worked example: Task, write a note to the team about the new meeting schedule. Audience, the whole department, including two part-time staff. Guardrails, no jargon, no promises about dates, stay under 150 words. That prompt takes ten seconds to write and returns something you can send with small edits.

TAG in your job

TAG is deliberately light on context. That keeps the prompt short and the answer focused, which suits small, contained decisions you can describe in one line. When a decision is big, add the background yourself, a few lines on history, constraints and who is affected, or move to RISEN, which has a dedicated slot for context.

The action line three options with trade-offs, how each could fail, then your recommendation is a reusable block. Save it and paste it into any prompt where you want a decision rather than a summary. A good Task names the decision, not the topic. If you cannot say it in one line, you are probably holding two decisions and should split them. The Goal line is what the AI optimises for. Change it and the recommendation changes.

TAG works the same way in ChatGPT, Claude and Gemini. The framework is yours, not the tool's, so you can carry it between apps without rewriting it. This week, write one TAG prompt for a real decision. Save the action line. Next time, you will not start from a blank page.

Frequently asked questions