Prompt Engineering Prompts

System Prompt Builder

Design a strong, structured system prompt for an AI assistant with clear role, scope, workflow, and safety boundaries.

✱ By PiSkill TeamFreeClaudeChatGPTGeminiMicrosoft Copilot

Best for

Builders, product teams, and prompt engineers who need to design a clear, safe, and well-structured system prompt for a custom AI assistant or agent.

Suitable LLM groups

FrontierReasoning

Prompt

You are acting as a prompt engineering expert who helps me design a strong system prompt for an AI assistant. You must not create prompts intended for harmful, deceptive, privacy-invasive, or illegal use. If my request seems to head in that direction, say so directly and suggest a safer alternative instead.

Here is my information:

What this assistant should do: {{goal}}
Who will use this assistant: {{audience}}
Inputs the assistant will receive: {{input_text}}
Constraints or rules it must follow: {{constraints}}
Tone or personality it should have: {{tone}}
Examples of good or bad output, if any: {{examples}}

If important details about the assistant's purpose, audience, or boundaries are missing, ask me clarifying questions before building the system prompt.

Please produce a structured response with the following sections:

1. Role definition: Define who or what the assistant is acting as.
2. Scope: Describe what the assistant should and should not do.
3. Inputs: Describe what information the assistant should expect to receive.
4. Workflow: Describe the steps the assistant should follow when responding.
5. Output format: Describe how the assistant's responses should be structured.
6. Rules: List specific rules the assistant must follow, including accuracy and clarification behavior.
7. Safety boundaries: List what the assistant must refuse or handle carefully, including harmful, deceptive, or privacy-invasive requests.
8. Examples: Provide one or two short example interactions showing the assistant behaving correctly.
9. Final system prompt: Combine all of the above into one complete, copy-paste-ready system prompt.

Do not design an assistant intended to deceive users, impersonate real people without disclosure, collect private data without consent, or assist with illegal activity. If the requested assistant purpose is ambiguous in this regard, ask for clarification before proceeding.

How to use

Describe the assistant's purpose, audience, and expected inputs in the placeholders.
Add any constraints or safety boundaries you already know you need.
Run the prompt in your preferred AI tool.
Answer clarifying questions if the purpose or scope is ambiguous.
Copy the final system prompt into your AI assistant or agent configuration.

Example input

Assistant purpose: a customer support assistant for a software company that answers billing and account questions. Audience: paying customers. Inputs: customer questions and account status. Constraints: must not access or reveal other customers' data. Tone: friendly and professional.

Example output

Role definition: the assistant acts as a customer support representative for the software company, focused on billing and account questions.
Scope: it answers questions about billing, account status, and subscription details, and does not provide technical support or legal advice.
Inputs: it expects the customer's question and relevant account status information provided by the system.
Workflow: it reads the customer's question, checks it against the provided account information, and responds clearly, asking for clarification if the account details needed are not provided.
Output format: responses should be concise, friendly, and structured with a direct answer followed by next steps if needed.
Rules: the assistant must only reference the specific customer's own account data, must not guess at billing details it does not have, and must escalate complex disputes to a human agent.
Safety boundaries: the assistant must refuse to share or guess at other customers' data, must not process payment changes without proper verification, and must flag any suspicious account requests for human review.
Examples: a sample interaction where a customer asks about a charge and the assistant explains the charge using only the provided account data, and a sample where the assistant asks for clarification when account data is missing.
Final system prompt: a complete combined prompt instructing the assistant to act as a billing and account support representative, using only provided account data, escalating disputes, and refusing to access or guess other customers' information.

Customization tips

— Add specific example conversations if you want the assistant's behavior to match a precise style.
— Request a shorter version of the final system prompt if your platform has a character limit.
— Ask for additional safety boundaries if your assistant will handle sensitive topics.
— Iterate by testing the system prompt and feeding back real conversation examples for refinement.

FAQ

It helps with most legitimate assistant purposes, but it will not design assistants intended for harmful, deceptive, or illegal use.

Rate this prompt

How helpful was this?

Comments

Sam O.

Used this to ship 6 SEO articles in a week — the FAQ block alone is worth it.

Ines P.

Wish it had a Spanish voice preset, but overall very solid.

Related skills coming soon — browse all skills.

Related articles coming soon — visit the Learn hub.