Universal Prompt Builder
Fill in a structured form and get a ready-to-copy AI prompt — role, task, format, tone, and constraints included.
A prompt built from separate role, task, format, and tone fields is more consistent than one written freeform, because each piece gets defined on its own instead of blending into a paragraph where details are easy to skip. This tool combines the four independently, so you can swap just one field — the tone, say — without rewriting the whole prompt.
How to use it
This is the actual thing you want the AI to do — be specific about the subject, not just the format.
A defined role, like "senior copywriter," shapes vocabulary and judgment calls the AI makes while completing the task.
Format controls structure — list, table, prose. Tone controls how it reads — formal, casual, direct.
Tips for better results
- Be as specific in the task field as you would be talking to a person. Vague tasks produce vague prompts — "write about marketing" generates a weaker prompt than a specific request naming an audience and channel.
- Regenerate with a different role before changing the task. The role field often has more influence on the final output's voice than small task wording changes do.
- Leave format on Any only if you genuinely don't care how the output is structured. A specific format chosen upfront avoids getting an unusable paragraph you have to reformat yourself afterward.
Example output
“Role: Senior UX writer. Task: write 3 microcopy variants for an empty search-results state. Format: bulleted list. Tone: friendly, brief.”
TL;DR
Feed an AI assistant a vague request and you get a vague answer back. Type “write me a prompt for a product description” into ChatGPT and you’ll get something usable eventually, but usually only after two or three follow-up messages clarifying what you actually meant.
Why a structured prompt beats a blank text box
Most weak prompts fail for the same handful of reasons: no defined role, no stated format, no length limit, no example of what “good” looks like. A model faced with “write about running shoes” has to guess whether you want a product description, a blog post, or a comparison chart — and it will guess, which is exactly the problem. Every guess it makes is a decision you didn’t get to make yourself.
This builder forces you through each of those decisions before you copy anything. That doesn’t mean every field needs an answer — plenty of solid prompts skip Context or Examples entirely — but it means the fields you skip are the ones you’re deliberately choosing to skip, not ones you forgot existed. That distinction is the entire difference between a prompt that needs three rounds of follow-up questions and one that works on the first attempt.
What a good prompt actually needs
Every field in this tool maps to a specific job a prompt has to do:
- Role sets the lens the model uses to answer. “You are a senior copywriter” pulls different vocabulary and structure than no role at all, or than “you are a career coach.”
- Task is the one field that actually can’t be skipped — it’s the instruction itself, and the more concrete the verb (“write,” “summarize,” “compare”), the less room there is for the model to wander.
- Context gives background the model has no way to infer on its own — who the audience is, what’s already been tried, what the output is for.
- Format decides the shape of the answer. Asking for “a bulleted list” versus leaving format open is often the single biggest lever on how usable the first draft is.
- Tone and length are smaller levers, but they’re the ones people forget to state and then complain about after the fact — “too formal,” “way too long.”
- Constraints are guardrails: a word count, a rule to avoid (“no exclamation points”), a hard requirement (“must include the price”). This is usually the highest-leverage field for a prompt that already has a role and task.
- Examples do more work than any instruction can. Showing one sample of the output you want communicates tone, structure, and length all at once, in a way a paragraph of instructions can’t match.
Example: from vague idea to structured prompt
Someone typing “write a product description for running shoes” directly into ChatGPT will usually get back a generic, adjective-heavy paragraph. Running the same idea through this builder — Task: “Write a product description for a running shoe,” Format: “a plain paragraph,” Tone: “direct,” Constraints: “under 60 words, no exclamation points, end with one concrete spec” — produces a prompt that reads:
“Write a product description for a running shoe. Format the response as a plain paragraph. Use a direct tone. Constraints: under 60 words, no exclamation points, end with one concrete spec (weight, drop height, or material).”
Same starting idea, but the second version tells the model exactly what “done” looks like — which is why it usually needs zero follow-up edits.
Who this is for
This tool is built for anyone who uses ChatGPT, Claude, or Gemini regularly and is tired of rewriting the same kind of prompt from memory every time — marketers drafting copy, students structuring study questions, developers writing prompts for documentation, or anyone who’s noticed their first message to an AI assistant rarely gets the result they actually wanted. It’s also useful as a teaching tool: watching which fields change the output most is one of the fastest ways to internalize what makes a prompt work.
FAQ
Do I need to fill in every field?
No. Only the Task field is required. Everything else is optional, but the more you fill in, the more specific the output — Constraints and Format tend to matter most.
Which AI tools does the generated prompt work with?
The output is plain text, so it works with ChatGPT, Claude, Gemini, or any other text-based AI assistant. It’s not tuned for image models — for those, use the Negative Prompt Generator alongside your own descriptive prompt.
Does this tool send my prompt anywhere?
No. Everything runs in your browser. Nothing you type is sent to a server or stored.
Can I save prompts I’ve built for reuse later?
Not yet inside the tool — copy them into your own notes for now. A save/history feature is on the roadmap.
Why does adding a role change the output so much?
A role narrows the model’s assumed vocabulary, structure, and level of expertise before it even reads the task. “You are a Python tutor” versus no role at all can be the difference between a beginner-friendly explanation and a terse technical one.
