About this tool
Good answers start with a clear prompt. Fill in simple fields — the role the AI should take, the task, the context it needs, the audience, tone, output format and rules — and get a well-structured prompt you can paste into ChatGPT, Claude, Gemini or use as a system prompt in an app. Start from ready templates for product descriptions, WhatsApp replies, document summaries, Google Ads copy and blog outlines.
How to use it
- Pick a template or start blank, and describe the task — the one field you must fill in.
- Add the role, context, audience, tone, format and any rules (one per line) that matter for your answer.
- Choose Markdown headings or XML tags, then copy the prompt into your AI tool.
Good to know
- Empty fields are left out, so short prompts stay short.
- Rules are turned into a bullet list, which models follow more reliably than a long paragraph.
- XML-style tags (<task>, <context>…) are a good choice for Claude and for system prompts in apps; Markdown headings work well everywhere.
- Templates contain [brackets] for you to replace with your own details.
- The prompt and its approximate token count are built in your browser; nothing is sent anywhere.
The parts of a strong prompt
Role sets the expertise and voice (“an experienced tax adviser”). Task says exactly what to produce. Context gives facts the model can't know — your prices, your policy, the document. Audience and tone decide how it should sound. Output format describes the shape of the answer, and rules set limits such as length or “only use the facts given”.
You don't need every part every time. A short task with clear context and a format usually beats a long prompt full of general advice.
Before and after
Weak: “Write about our sofa.” Strong: role — an e-commerce copywriter; task — write a product description for the Karachi 3-seater sofa; context — solid sheesham wood frame, washable covers, delivery in Karachi within 5 days; format — one hook line and four bullets; rules — under 100 words, no invented features. The second prompt gets a usable answer first time.
Using it as a system prompt
When building a chatbot or AI feature, put the role, rules and output format in the system prompt and send the user's message separately. The XML style is handy here: tags such as <rules> and <output_format> make it easy for the model — and for your team — to see where each instruction lives. Check the length with the AI Token Counter, because the system prompt is sent with every request.
Frequently asked questions
What makes a good AI prompt?
A specific task, the context the model can't guess, who the answer is for, the format you want back and any rules — such as a word limit or “only use the facts given”. This builder asks for each of these.
Should I use Markdown or XML tags?
Both work. XML tags make it very clear where each part starts and ends, which helps with long context and system prompts; Markdown headings are easier to read and edit.
What is a system prompt?
Instructions given to a model before the conversation starts — usually the role, rules and output format for an app or chatbot. A prompt built here can be used as a system prompt as it is.
How do I stop the AI from making things up?
Give it the facts in the Context field, add a rule such as “Only use the information above and say ‘not stated’ if something is missing”, and tick the option to ask clarifying questions.