ChatGPT Prompts: Ready-Made Templates You Can Copy
In short
A prompt is the text instruction you hand to an AI model. A good ChatGPT prompt almost always has four parts: role, task, context, and output format. Below is a library of ready-made templates for work, writing, studying, and everyday life. Copy them straight out of the code blocks and drop your own details into the square brackets.
We'll also cover how prompting for ChatGPT differs from prompting Claude, how to adapt any template to your own case, and what you need to know about ChatGPT access from Russia (as of July 2026). If you want a wider set, see the main pillar on ready-made AI prompts and the prompt arsenal.
What makes a good ChatGPT prompt?
Prompting is the skill of phrasing a request so the model gets the task right the first time. ChatGPT is decent at guessing intent, but answer quality jumps sharply once you set the boundaries.
A prompt structure that works:
- Role — who the model should be ("you are an editor", "you are a math tutor").
- Task — exactly what to do, in one verb ("write", "shorten", "compare").
- Context — facts, numbers, constraints. The more specific, the less filler.
- Format — how to deliver the result: table, list, under N words, with an example.
One honest caveat: ChatGPT can confidently invent facts (these are called hallucinations). That's why it helps to build a safety catch into your prompts — something like "don't make anything up; if you don't have the data, say so." For a deeper dive into the technique, see the guide on how to write prompts.
And one more rule for beginners: don't try to get the perfect answer on the first try. ChatGPT works well in dialogue — refine as you go ("make it shorter", "add an example with numbers", "too formal, soften the tone") and the result converges fast. That's usually quicker than polishing one long prompt.
Ready-made prompts for work
A business email:
You are a business assistant. Write an email to a client.
Context: we are missing the delivery deadline by 3 days because of a customs delay.
Tone: polite but confident, no half-page apology.
Length: under 120 words. Propose a new date and a small compensation.
A meeting summary from a transcript:
Here is a meeting transcript: [paste text].
Produce three blocks: 1) a 5-bullet summary,
2) decisions made, 3) action items as "who — what — by when".
If a deadline was not stated, write "deadline not specified" — don't invent one.
Breaking a task down:
Help me break a task into steps.
Task: launch a landing page in a week, working solo.
Give me a day-by-day plan, flag what can be delegated or simplified,
and where I'm most likely to get stuck.
Prompts for writing, studying, and daily life
Rewriting and simplifying text:
Rewrite this text to be simpler and shorter; keep the meaning and every number.
Strip out the bureaucratic phrasing, break it into short paragraphs.
Reader level: high school senior. Text: [paste].
Explaining a topic for study:
Explain the topic "compound interest" as if to a beginner.
Give me: a one-sentence definition, a real-life example with numbers,
a common mistake, and one mini-exercise with the answer.
A weekly plan for the household:
Build a 5-day meal plan.
Budget: tight. Cooking time: under 30 minutes. No seafood.
Format: a "day — breakfast — lunch — dinner" table plus a shopping list.
Brainstorming ideas:
Give me 10 headline ideas for a post about saving money on subscriptions.
Vary the tone: 3 serious, 3 funny, 4 with numbers.
Keep them short — 8 words max each, no clickbait, no ALL CAPS.
These templates are deliberately simple: swap the text in the square brackets and you get a usable result. More ready-made starting points for specific roles and scenarios live in the prompt arsenal.
How does prompting for ChatGPT differ from Claude?
The basic principles are the same, but the nuances differ. Here's an honest comparison, marketing aside.
| Aspect | ChatGPT (GPT-5.x) | Claude |
|---|---|---|
| Long instructions | Likes lists and explicit structure | Holds long context and multi-step rules well |
| Role up front | Helps, but not critical | Often answers more precisely with an explicit role |
| Output format | Ask for the format directly (table, JSON) | Same; follows strict formats carefully |
| Code and agents | Strong at quick snippets | The basis for Claude Code and coding agents |
| Default style | Friendly, sometimes padded | Drier, less filler |
The practical takeaway: the same template usually transfers between models, but for Claude it pays to spell out the role and the rules more firmly. A Claude-specific breakdown is in the article on prompts for Claude.
Common beginner mistakes
Even with a ready-made template it's easy to trip on the same things. A short checklist to run before you hit send:
- Too vague a request. "Write a post" tells the model nothing about topic, length, or audience. Add even one or two constraints and the answer sharpens immediately.
- No output format. Without an explicit "as a table" or "a list of up to 7 points", ChatGPT returns a wall of text you then have to sort out by hand.
- Five tasks in one message. Unrelated tasks in a single prompt drag down the quality of each — split them into steps and go one at a time.
- No guard against invention. If you're asking for facts or numbers, add "if you're not sure, say so" — otherwise you risk a plausible but false detail.
- Not reading the first answer through. One clarifying line often beats a brand-new prompt: "cut it in half" or "add an example" works faster than rewriting.
A separate note on safety: never paste passwords, clients' personal data, or trade secrets into the chat. Anonymize your examples — use placeholder names and amounts instead of real ones — and move the finished answer into your own document.
How do you adapt any template to your own task?
Take any prompt above and run it through the role → task → context → format formula. Replace the generic wording with your own numbers and constraints — context is what delivers most of the answer quality.
Once a prompt stops being a one-off message and becomes a working tool, single phrases aren't enough. Vibe coding is the approach where you assemble a working product by describing it to an AI in words instead of writing code by hand. For that kind of work in Claude Code or Cursor you need not one prompt but a connected system of proven building blocks and steps.
That's exactly what the qvib engine does: it turns an idea into a project through structured steps and removes the blank-page fear. And if you're still getting comfortable with phrasing, start with the prompt arsenal and the guides, then carry what you learn into a real project.
ChatGPT access from Russia: the facts, no myths
Let's look at the situation honestly, as of July 2026. Officially, ChatGPT is not available in Russia: OpenAI restricts access from Russian IP addresses and does not accept cards issued in Russia (including Mir, the domestic payment system, and Visa/Mastercard from Russian banks), and it does not price in rubles.
What that means in practice:
- Price. The official ChatGPT Plus subscription is $20 per month (as of July 2026). Converted to rubles, with fees and exchange rate, that usually lands around 2,000–2,500 ₽ — the exact amount depends on how you pay.
- Free tier. It works, but with hard message limits — once you hit them you wait several hours. For the templates in this article, the free version is usually enough.
- Payment. The legal options are a foreign (virtual) card, with the caveat that reliability isn't guaranteed, or an aggregator service. An aggregator is an intermediary that gives you access to the model and accepts payment in rubles.
You won't find step-by-step ways around geo-blocks here — that's a separate topic. Legal alternatives and aggregators are covered in ChatGPT alternatives in Russia and how to use ChatGPT in Russia.
FAQ
Do I need paid ChatGPT to use these prompts?
No. Every template works on the free tier — there are just limits on the number of messages. A paid subscription ($20 per month as of July 2026) gives higher limits and access to the top models, but for emails, rewrites, and study explanations it isn't required.
Why does ChatGPT sometimes answer worse in Russian than in English?
The model was trained mostly on English-language text, so in Russian you occasionally get calques and clumsy phrasing. It helps to set the role explicitly ("you are a native Russian speaker"), ask for "no anglicisms", and specify the style. Quality improves year over year, but you still need to check the facts.
Can I use these templates in Claude or other AI models?
Yes, most prompts transfer with almost no changes. For Claude it's worth spelling out the role and format rules more firmly — details in the article on prompts for Claude. The underlying logic of building requests is the same across all modern chat models.
How do I pay for ChatGPT Plus from Russia?
OpenAI does not accept Russian cards. As of July 2026 the legal routes are a foreign or virtual card (with the caveat that it can stop working) or an aggregator that takes payment in rubles. Details and a comparison are in ChatGPT alternatives in Russia.