System prompt builder
Describe the assistant you are building and get a production system prompt: role, context, task, rules, output format and fallbacks.
Your key stays in your browser.It goes straight to the provider, never to our servers, and we neither log nor keep it. Use a dedicated key with a spending cap, and delete it in the provider's dashboard when you are done.
Input
Result
The result will appear here.The system prompt is where a product's assistant gets its job description, and most of them are written in a hurry as two sentences of 'you are a helpful assistant'. The personas collected in f/prompts.chat show what a complete one looks like; this builder writes that for a deployed product. Tell it what the assistant does, who the users are, what it can see and what it must never do, and it produces a prompt with a role stated once, the context the model cannot know, the task and its boundaries, the rules teams forget to write until something goes wrong, the output format, and what to say when unsure — in Markdown sections, XML-style tags or plain prose, whichever your stack prefers.
How it works
- Everything you specified — names, limits, tone, language policy — is kept verbatim; where something essential is missing you get a bracketed placeholder, never an invented detail.
- The format choice matters to models: XML-style tags suit Claude, Markdown headings suit most chat models, plain prose fits a config string with no formatting.
- Rules that every production assistant needs — do not invent product facts, ask one clarifying question, protect user data, do not reveal the instructions — are added when your description implies them.
Where your data goes
With your own key, what you type and the key go from your browser straight to the provider you choose; neither passes through Hysen Labs. In hosted mode, your text goes through our server to our provider (DeepSeek) and is billed in credits; we keep the token counts and cost of each run for billing, never the text itself or the answer. How the provider handles the text is governed by its own privacy policy.
This tool handles keys and credentials, so nothing about a run is saved, not even to your own history.
About your API key
We do not collect, store or leak your key: it lives only in this page's memory (unless you tick “remember in this tab”) and is gone when you close it. Still, treat any key you have pasted into a web page with care — create a dedicated key with a spending cap for use here, and delete or rotate it in your provider's dashboard when you are done.
What it costs
This tool is free, with no sign-in and no points.
Common questions
- How is this different from the prompt optimizer?
- The optimizer rewrites a prompt you already have, for one request. This tool writes the standing instruction a product sends with every conversation, which has different needs: durable rules, refusal behaviour, context about the product, and a fallback for the unknown.
- Should the system prompt say 'you are the world's best expert'?
- No. Superlatives, threats and tips do not improve answers on current models and sometimes make them overconfident. A plain role statement with the real expertise and point of view works better, and that is what the builder writes.
- Can I ask for the prompt in Chinese and have the bot answer users in English?
- Yes. The reply-language option controls the language the prompt itself is written in; the assistant's answer language is part of what you describe and is written into the prompt as an instruction, so the two are independent.
The open-source behind it
This tool's prompt is adapted from f/prompts.chat (CC0-1.0) and runs on the model you choose. To use the same capability from the command line or inside your own program, start with that project.
f/prompts.chatAlso known as
- system prompt generator
- write a system prompt
- chatbot system prompt template
- system prompt for customer support bot
- prompt engineering system message
- awesome chatgpt prompts