Assemble a professional system prompt from roles, tones, constraints and output formats. Copy the finished prompt and use it in ChatGPT, Claude, Gemini or any LLM.
A system prompt sets the rules the AI follows for the whole conversation: who it is, how it talks, what it may or may not do. A good one makes answers more consistent, accurate and on-format — a vague one leaves the model guessing.
The builder runs entirely in your browser. Your prompt and text are never sent to any server and never stored.
It is the opening instruction that defines how an AI behaves for the whole conversation: its role, tone, rules and output format. It is supported by ChatGPT (Custom instructions), Claude, Gemini and most LLM APIs.
No. It assembles the prompt from your choices — no API, no cost, no waiting. The finished text is designed to work well when you paste it into ChatGPT, Claude, Gemini or any other model.
In ChatGPT use Custom Instructions (or the Project instructions), in Claude use System Prompt in settings or the API, in Gemini use the system instruction field. For APIs, it goes in the system message.
Clarity and specificity: a clear role, a concrete task, the desired output format, and rules about what not to do (e.g. never invent facts, ask when unsure). Short, direct instructions work better than long vague ones.
Yes, with small adjustments. Most models follow the same instruction style. Very large context models handle longer prompts, while smaller ones respond better to the shortest possible version.
Yes. Everything is built locally in your browser — nothing is uploaded, logged or stored.