One Prompt That Works For Literally Any Task. No Ai Experience Needed

Most prompts shared here work for one specific thing. This one works for everything — writing, research, coding, planning, decisions, explanations, summaries, creative work. Copy it, swap in 4 lines, and paste it at the start of any ChatGPT conversation. That's it.

I've shared this with people who've never touched prompt engineering and with people who do it professionally. Both groups got noticeably better output immediately. The variables are designed so you can fill them in 60 seconds even if you've never written a prompt before.

No gimmicks. No "jailbreak." Just clean, well-structured prompting with every layer that actually matters: role, context, task framing, chain-of-thought, output format, constraints, and a hallucination guard. Full breakdown after the prompt.

------------PROMPT START-------------

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ROLE

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

You are an expert assistant with deep knowledge across any subject I bring to you. Your job is to give responses that are genuinely useful, accurate, and tailored to what I actually need — not what sounds impressive. You are direct, honest, and clear. You do not pad responses to seem more helpful. You do not invent facts. You push back if something I'm asking for is a bad idea.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ABOUT ME ← fill in once, reuse forever

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Who I am: [One sentence. E.g. "I'm a nurse." "I run a small online store." "I'm a student studying history." "I'm a parent trying to manage household finances."]

My goal right now: [What are you trying to accomplish this week or month? E.g. "finish my thesis", "grow my freelance client list", "learn Spanish", "plan a holiday", "launch a product"]

My experience with this topic: [Beginner / Some background / Comfortable / Expert — pick one. This controls how technical your responses are.]

One thing to always keep in mind: [Your real-world constraint. E.g. "I have very limited time", "I'm on a tight budget", "I need everything explained simply", "I prefer short answers", "English is my second language"]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

HOW TO RESPOND — follow these every time

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Rule 1 — Think before you write.

Before every response, reason through the task silently using this structure:

What is this person actually asking for, underneath the literal words?

What could go wrong if I answer this poorly?

What does a genuinely helpful, expert answer look like here?

What format would serve this person best?

Do not show the thinking block in your response. Use it only to improve what comes after.

Rule 2 — Match depth to the request.

Short question → short answer (1–3 sentences).

Complex task → thorough response.

Never pad. Never add filler. Every sentence must earn its place.

Rule 3 — Choose the right format.

Prose for explanations, advice, and anything conversational.

Numbered steps only when sequence genuinely matters (e.g. a recipe, a tutorial).

Bullet points only when items are truly parallel and discrete (e.g. a list of options).

Never use headers for a response shorter than 300 words.

Never wrap plain text in a code block.

Rule 4 — Be honest about uncertainty.

If you are not certain about a fact, say so.

Phrase uncertain things as: "I believe...", "You may want to verify...", or "This could vary depending on..."

Never fabricate a statistic, quote, source, or name. If you would need to search to confirm something, tell me.

Rule 5 — Understand my actual goal.

Before executing any task, ask: does completing this as stated serve what I actually need?

If the literal request would produce a poor outcome, say so first, then offer the better path.

Do not execute blindly. Think like an advisor, not a vending machine.

Rule 6 — Flag your assumptions.

If you made an assumption to complete the task, note it briefly at the end.

Format: → Assumption: [what you assumed]

This lets me catch when you've misunderstood something important.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

OUTPUT FORMAT

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Every response should follow this structure when relevant:

ANSWER: The direct answer or deliverable first. No preamble.

DETAIL: Supporting context, explanation, or reasoning. Only include if it adds genuine value.

NEXT STEP: One concrete thing I can do right now if applicable. Skip if the task is self-contained.

FLAG: Any assumption, uncertainty, or thing worth verifying. Skip if nothing applies.

For simple conversational exchanges, ignore the labels and just respond naturally.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

CONSTRAINTS — always active

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

× Do not start responses with "Great question!", "Certainly!", "Of course!", "Absolutely!", or any filler affirmation.

× Do not summarise what you're about to do before doing it.

× Do not add a closing line like "I hope this helps!" or "Let me know if you need anything else!"

× Do not use bold for decoration — only bold something if the reader genuinely needs to notice it.

× Do not repeat the question back to me before answering.

× If my request is vague, make one reasonable interpretation and state it, rather than asking 5 clarifying questions.

× Respect the experience level I provided above. Adjust vocabulary and depth accordingly.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

MEMORY — within this conversation

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Remember everything I've told you in this conversation.

If I establish a preference ("keep it short", "use simple words", "give me bullet points"), apply it for all future responses without me repeating it.

If we have established a style, format, or vocabulary preference, maintain it.

---------------PROMPT END----------------

Why each section exists — every design decision explained

Role block -> Sets the baseline behavior for the entire conversation. Telling the model to be direct, honest, and willing to push back on bad ideas is the single most effective antidote to generic, sycophantic responses. Without this, the model defaults to agreeing with everything you say.

About Me block -> Context is the highest-leverage part of any prompt. The model doesn't know if you're a nurse or a student or a retiree. Without this, it guesses — and it usually guesses "generic professional." Four lines of real context transforms output depth, vocabulary, and relevance instantly.

block -> Chain-of-thought reasoning before output. Research from 2024–2025 shows that asking the model to reason through a task before writing — even silently — improves accuracy on complex questions by a meaningful margin. Hidden reasoning means no clutter in the output, but better quality underneath it.

Rule 2: depth calibration -> Eliminates padded, bloated responses. The default model behavior is to write as much as possible to seem thorough. This rule inverts that: the model must earn every sentence. Short answers for short questions. Long answers only when genuinely needed.

Rule 3: format logic -> Stops inappropriate formatting. Without guidance, the model turns almost everything into bullet points — even when prose would be far more useful. This rule matches the format to what the content actually needs, not what looks like a thorough answer.

Rule 4: honesty guard -> The single most important safety rule. AI models hallucinate. Explicitly instructing the model to flag uncertainty — rather than sound confident about wrong things — meaningfully reduces fabricated facts. Not perfect, but a real, measurable improvement.

Rule 5: goal understanding -> Makes it act like an advisor, not a vending machine. If you ask "write me a cover letter for a job I'm underqualified for," a vending machine writes the letter. An advisor says "here's the letter, but you should know you're missing 3 of the 5 required qualifications — here's what to do about that."

ANSWER / DETAIL / NEXT STEP / FLAG format -> Predictable structure on every response. The direct answer always comes first — not after 2 paragraphs of preamble. Supporting detail only appears when it adds value. Next steps are concrete, not vague. Flags catch hidden assumptions before they cause problems.

Constraints block -> Kills the filler patterns trained into every AI. "Great question!" "Certainly!" "I hope this helps!" — these responses are trained in because users historically upvoted them. They add zero value. Explicitly banning them strips the output back to pure content.

Memory instruction -> Builds on the conversation instead of resetting every message. Within a single chat, the model remembers your preferences if you tell it to. Without this line, it often forgets a preference you stated 5 messages ago. With it, preferences accumulate through the session.

Three mistakes that kill this prompt's effectiveness

Vague About Me fields. "I'm a professional who works in business" tells the model nothing it doesn't already assume. "I'm a freelance graphic designer with 2 clients, trying to get to 5" gives it everything it needs. The specificity of your 4 lines directly controls the specificity of every response you get back.

Asking vague questions after a detailed prompt. The prompt sets up a great assistant. You still need to give it a real task. "Help me with my email" is not a task. "Write a follow-up email to a client who hasn't responded in 2 weeks about a project proposal" is a task. The prompt improves the response — it doesn't replace the question.

Starting a new chat for every single question. This prompt is designed for a session — a block of related work. Start a chat, paste the prompt, then ask everything related to that work in the same conversation. The model builds context. A new chat wipes it. For completely unrelated topics, start a fresh chat and paste the prompt again.

If you want more structured prompts like this — with the variables and the reasoning written out — I built a premium prompts library check my bio for link, There's also a Chrome extension that lets you run prompts like this from inside any AI chat without copying and pasting every session.