The CO-STAR Framework
Context, Objective, Style, Tone, Audience, Response — six pieces that turn a rough request into a message that actually lands with its reader. CO-STAR is the framework to reach for when who you are writing to and how it sounds matter most, whether that is an email to a professor or an announcement to a class.
What is the CO-STAR framework?
The CO-STAR framework is a simple checklist for writing better prompts when the goal is communication. Instead of asking the AI to "write an email" and hoping it strikes the right note, you give it six things it almost always needs: the Context behind the message, the Objective you want it to achieve, the Style of writing, the emotional Tone, the exact Audience who will read it, and the Response format you need back.
Large language models like ChatGPT, Claude, and Gemini are not mind readers. They generate text by predicting what should come next based on what you gave them — so when your prompt does not say who the message is for or how it should sound, the model defaults to a bland, one-size-fits-all voice that fits no one in particular. CO-STAR works because it removes that guesswork. Naming the audience and the tone is the difference between an email that reads as respectful and one that accidentally sounds entitled. (For the bigger picture on why this matters, see why better prompts matter.)
CO-STAR is particularly useful for students and educators, because so much of school life is communication that has to be judged just right. A student emailing a professor to ask for an extension needs to sound polite and accountable, not casual or demanding; a teacher announcing a new project to families needs to be clear, warm, and reassuring all at once. The rest of this guide breaks down each element with examples drawn from real messages students and educators send every week.
The six elements of CO-STAR
Each card shows the question the element answers, with examples for studying, teaching, and everyday communication.
Context (C)
What background should the AI know about the situation?
- •I missed two labs because I was in the hospital this week.
- •Our spring field trip has been moved to a new date.
- •A professor wrote me a strong recommendation last year.
Objective (O)
What is the message actually trying to achieve?
- •To request a short extension and propose a new due date.
- •To ask a professor whether they will write me a reference.
- •To get parents to sign and return a permission slip on time.
Style (S)
What kind of writing should it be?
- •Formal academic email with a clear subject line.
- •Plain, jargon-free language any parent can follow.
- •Short and scannable, with key dates in bold.
Tone (T)
How should the message feel to the reader?
- •Polite, respectful, and accountable — not making excuses.
- •Warm and reassuring, but still clear about the deadline.
- •Encouraging and supportive of a struggling student.
Audience (A)
Who exactly will read this?
- •A professor I do not know well and want to impress.
- •Busy parents who skim email on their phones.
- •Ninth-graders who are nervous about a big assignment.
Response (R)
What format do you need back?
- •A short email under 150 words with a subject line.
- •A class announcement of three short paragraphs.
- •Two versions so I can pick the one that fits best.
Style vs. Tone?
These two are easy to confuse. Style is the kind of writing — formal email, bulleted list, casual note. Tone is the feeling behind it — warm, firm, apologetic, encouraging. You can keep the same formal style while shifting the tone from celebratory to consoling, and naming both gives you precise control over how the message comes across.
Why CO-STAR produces better answers
Each element of CO-STAR fixes a specific failure mode in written communication. Context stops the model from inventing details or sounding generic — it is the difference between a vague "please excuse my absence" and a message that references the actual situation. Objective keeps the message focused on one clear ask, so the reader knows exactly what you want them to do.
Style and Tone are where CO-STAR really earns its keep. Together they control how the message feels: the same request can read as entitled or as gracious depending on a few word choices, and a generic AI default tends to land somewhere bland and slightly off. Audience is the most powerful lever of all — telling the model it is writing to a professor versus a classmate versus a parent reshapes the vocabulary, the formality, and the level of explanation automatically. Response then makes sure you get back something you can actually send, instead of a draft you have to reformat.
You do not need to label the parts with the literal words "Audience" and "Tone" — the model does not require them — but writing them out keeps you honest and makes sure nothing important is missing before you hit send. That is the real value of a framework: it is a thinking aid as much as a formatting one. If you would rather not assemble the pieces by hand, the CO-STAR prompt generator gives you a field for each element and writes the finished prompt for you.
CO-STAR in action: before and after
The same message, written two ways. Notice how much the model has to guess in the first version.
Write an email to my professor asking for an extension on my lab report.
[Objective] Politely ask Professor Lin for a three-day extension and propose Monday as a new due date, while showing I take responsibility for the work.
[Style] A short, formal email with a clear subject line and a proper greeting and sign-off.
[Tone] Respectful and accountable — apologetic without over-apologizing, and not making it sound like I expect a yes.
[Audience] A professor I do not know well yet, who is busy and values students who are direct and prepared.
[Response] One email under 150 words, ready to send.
The vague version produces an email that sounds like every other extension request and risks coming across as entitled. The CO-STAR version reads as a considerate, self-aware student who is owning the situation and offering a concrete plan — exactly the impression you want to make on a professor you may need a recommendation from later.
Write an email to parents about the science fair project.
[Objective] Explain the project, list the key dates, and reassure families that no expensive supplies or special expertise are required.
[Style] Plain language with the important dates in bold, easy to skim on a phone.
[Tone] Warm and encouraging, but clear and organized so nothing gets missed.
[Audience] Busy parents and guardians with a wide range of science backgrounds, some of whom may worry they cannot help.
[Response] A short email of three paragraphs with a friendly subject line.
Because the prompt names the audience (busy parents, some anxious about helping) and the tone (warm but clear), the AI produces an announcement that informs without overwhelming — reassuring families up front, highlighting the dates they need, and lowering the barrier to participation rather than reading like a list of requirements.
Common mistakes to avoid
- Skipping the Audience. The audience is what makes the message fit the reader instead of sounding generic. "Writing to a professor I have never met" produces a very different email than "writing to my lab partner" — and it is the single most valuable thing you can specify.
- Treating Style and Tone as the same thing. Style is the format and register; tone is the emotion. You can write a formal email (style) that still feels warm and reassuring (tone). Setting both gives you control a single instruction cannot.
- Leaving out the Objective. If the model does not know what you want the reader to do, the message wanders. State the one clear ask — sign the form, grant the extension, reply by Friday.
- Asking for "professional" and nothing else. "Professional" tells the model almost nothing. "Respectful and accountable, but not groveling" or "warm but firm about the deadline" actually shapes the wording.
- Forgetting to give it the real details. CO-STAR structures your request; the AI still needs the facts — names, dates, the actual situation. Paste those in, then read the draft before you send it.
When to use CO-STAR (and when not to)
CO-STAR is the right tool when the message is going to be read by another person and the audience and tone matter — emails, announcements, requests, apologies, and persuasive or sensitive notes. For other situations, one of the sister frameworks may fit better:
- →CRAFT (Context, Role, Action, Format, Task) — for substantial, well-structured documents like study guides, essays, and lesson plans.
- →RISE (Role, Input, Steps, Expectation) — for multi-step assignments where you want the AI to follow a procedure.
- →CARE — when you need the output in a strict, repeatable format every time.
- →TAG (Task, Action, Goal) — for quick, everyday, one-off requests where CO-STAR would be overkill.
Not sure which to pick? How it works walks through the whole methodology, and the use cases page shows the frameworks applied to real situations.
Frequently asked questions
What does CO-STAR stand for in prompt engineering?
CO-STAR stands for Context, Objective, Style, Tone, Audience, and Response. It is a structured prompt framework built for communication: you tell the AI the background, the goal of the message, the writing style and emotional tone you want, exactly who will read it, and the format you need back — so the response lands correctly with its intended reader on the first try.
Is the CO-STAR framework good for students?
Yes. CO-STAR is ideal whenever a student has to write something that another person will read and judge — an email to a professor, a message to a group-project teammate, or a request for a recommendation letter. Naming the audience and the tone keeps the message polite and well-pitched instead of either too casual or stiff and awkward.
Do I have to use all six CO-STAR elements every time?
No. CO-STAR is a checklist, not a rulebook. For a routine message you might only set the Audience, Tone, and Objective. The more sensitive or high-stakes the message — an apology, a difficult announcement, a request from someone with authority over you — the more every element earns its place. For a quick, low-stakes task, the simpler TAG framework is a better fit.
Which AI models does the CO-STAR framework work with?
All of them. CO-STAR is model-agnostic — the same structured prompt improves results in ChatGPT, Claude, Gemini, Microsoft Copilot, and open-source models, because every large language model writes a better-targeted message when you spell out the audience, the tone, and the goal instead of leaving it to guess.
How is CO-STAR different from the CRAFT framework?
Both are structured frameworks, but they emphasize different things. CRAFT highlights Role and Format, which makes it strong for detailed, well-organized documents like study guides, essays, and lesson plans. CO-STAR instead adds an explicit Audience and Tone, which makes it the better choice when the reader and the voice of the message matter most — emails, announcements, and persuasive or sensitive writing.
Ready to try it out?
Open the builder, fill in the six fields, and copy your finished prompt into ChatGPT, Claude, or Gemini.