Writing Prompt Frameworks: CREATE, ROSES, STOKE, TRACE and More

Study guide

Write Detailed Prompts With Seven Writing Frameworks

Quick prompts produce bland text. When you ask an AI to write a blog post or a report with no character, no example and no constraints, it produces the safest average text it can. This course teaches seven frameworks that fix that: CREATE, ROSES, STOKE, TRACE, RELIC, BLOG and 4S. Each one is a fixed set of slots you fill before you send the prompt, and each slot names a decision the AI would otherwise guess, such as voice, audience, structure or length.

The course is built for writers, marketers and experts who produce detailed or long form content, and for anyone who pastes prompts into ChatGPT, Claude or Gemini and wants a usable draft on the first attempt. You will see how each framework works, when to choose it, and how to combine it with CO-STAR, which carries audience and tone while the writing framework carries the writing itself. Every lesson includes a worked example from a real job, from a pillar blog article to API documentation to a press pitch.

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 Writing frameworks, which is a fill-in-the-slots framework: an acronym where each letter stands for one part of your prompt. It is best for: writers, marketers and experts producing detailed or long-form content.

Why detailed writing frameworks beat quick prompts

A detailed writing framework is a fixed set of slots you fill before sending the prompt. Each slot names a decision the AI would otherwise guess, such as voice, audience or length. Bland output is usually a briefing problem, not a model problem. The AI had no character, no example and no constraints to work from, so it produced the safest average text. When the brief is thin, the draft is thin, no matter which model you use.

This course covers seven frameworks: CREATE, ROSES, STOKE, TRACE, RELIC, BLOG and 4S. They overlap in places, so treat them as different tool shapes rather than competing systems. Every one of them can be combined with CO-STAR. CO-STAR carries audience and tone, while the writing framework carries the writing itself, so the two sit side by side in one prompt.

A worked example helps. A content strategist briefing a pillar blog article names voice, audience, structure, length, one example and what to avoid. The reply then opens in the right voice without a second attempt. Slots only work when each one holds a real decision. Vague words in the slots produce vague text out, so the skill is in what you put in, not in the template itself.

CREATE: the most detailed writing framework

CREATE stands for Character, Request, Examples, Adjustments, Type, Extras. It is the most detailed of the common writing frameworks, which makes it a good fit when the output needs to be close to finished rather than a rough starting point. The slot order matters. Character sets the voice, Request states the task, Examples show the style, and the remaining slots refine. Working through them in that order keeps the prompt readable and the instructions connected.

The Examples slot is what sets CREATE apart. Instead of describing the style in words, you paste a sample of writing you already like, and the AI copies its rhythm, sentence length and tone directly. Adjustments hold the constraints, such as avoid cliches or keep it under a set length. They shape the draft rather than adding to it, so they belong after the Examples slot where the AI can read them against a real sample.

Extras hold the additions, such as include a call to action. They are easy to forget, and leaving them out is a common reason a draft feels unfinished even when the writing itself is good. The main failure mode is filler. A vague Examples slot or an Adjustments slot that says make it good gives the AI nothing concrete, so the extra length of CREATE buys you nothing over a shorter framework.

ROSES and STOKE: voice, structure and expert knowledge

ROSES has two versions. Version one is Role, Objective, Style, Example, Scenario. Version two is Role, Objective, Scenario, Expected solution, Steps. Both begin with Role and Objective, so you can switch versions without rewriting the start. Always spell ROSES out in the prompt. Writing only the letters gives the AI no instruction. Naming each slot also lets you check that nothing is missing before you send.

Use ROSES version one when voice matters, because Style and Example shape the tone. Use version two when structure matters, because Expected solution and Steps shape the answer. STOKE stands for Situation, Task, Objective, Knowledge, Examples. It is designed for domain expert writing, where the value comes from what you know, not from general knowledge the AI already has.

The Knowledge slot is where you paste your expertise so the AI does not fake it. Add your own data, methods and findings. Without it, the AI fills gaps with plausible but generic claims. A worked example shows the difference. A consultant writing a thought leadership article pastes client findings into Knowledge, and the reply builds its argument from those findings rather than inventing statistics.

TRACE and RELIC: making a brief testable

TRACE stands for Task, Requirements, Audience, Context and Examples. It suits technical writing and development briefs because each slot carries one clear job. The Requirements slot is where testable rules live. Write rules you can tick off, such as a method, a path and a sample response for every endpoint. Requirements and limitations make the brief testable. If you cannot say whether the answer passed or failed, the brief is still too vague.

RELIC stands for Role, Emphasis, Limitation, Information and Challenge. It adds emphasis and a challenge to push quality beyond a first acceptable draft. The Challenge step asks the AI to stress test its own output. In the API example, it named the endpoint with thin error cases, which is exactly the kind of gap a reviewer would find.

A weak brief says write some API documentation. A strong brief names the audience, the required sections and a length limit, so the output can be graded. That difference is the whole point of TRACE and RELIC. They turn a request into a set of rules you can check, and then ask the AI to look for where its own draft falls short.

BLOG and 4S for long form writing

BLOG stands for Background, Logic, Outline, Goal. The framework's key move is asking for the logic and outline before the article, so you can fix the structure before any long draft exists. Long form works best outline first. A wrong outline wastes thousands of words, while a wrong sentence costs one line. Approve the shape, then let the AI write.

4S stands for Structure, Style, Substance, Speed. It works as a checklist: name all four in the prompt so the AI knows the shape, the voice, the depth and the deadline. Substance and Speed pull against each other. Naming the trade in your prompt, such as depth first with a two day turnaround, gives a more useful draft than asking for high quality and quick.

BLOG and 4S combine well. Use BLOG to get the logic and outline, then add 4S lines for structure, style, substance and speed before the full draft. This approach connects to the decomposition course, where a big task is broken into approved steps. BLOG is decomposition applied to a single article.

Choosing a framework and finishing with a critique pass

The match map is the core of this lesson. Expert content calls for STOKE, creative content with samples calls for CREATE, technical writing calls for TRACE, and long form calls for BLOG. Choosing well takes seconds once you know the job. Picking a framework by fashion is a common trap. A framework chosen because it sounds impressive, rather than because it fits the job, wastes the structure and produces a mismatched draft.

Every framework needs your real knowledge and examples. The acronym organises what you already know. It cannot invent expertise, so a thin brief produces a thin draft no matter which framework you pick. A worked example shows a PR manager using STOKE to pitch a feature to a journalist. The subject, tone, outline, key facts and examples slots are all filled with real material, which is why the draft reads as newsworthy.

The critique pass is the final habit. After the AI drafts, switch roles and become a demanding editor. Ask what a sceptical reader would challenge and where the evidence is thin, then ask for a tighter version. Seven frameworks, one decision habit: choose by job, supply your expertise, and critique before publishing. That sequence works for any writing task you face.

Frequently asked questions