Custom AI Agent Builder
Generates a complete, production-ready system prompt for any agent you describe — with its role, operating procedure, refusal rules, stop conditions and a worked example, ready to paste.
Free — no card needed
Create a free account
to open Custom AI Agent Builder — and the other 364 prompts across 21 categories.
Already have an account?
Running it, start to finish
- Describe what the agent does and what it must never do.
- Paste the generated prompt and run the two suggested tests.
- Correct any assumption in the setup notes that does not match your case.
What you get back
The output this produces, every time.
- Produces a complete pasteable system prompt with the scope boundary, refusal rules and stop conditions most drafts omit.
- Includes a worked example in the prompt itself, which does more for output consistency than added instruction.
- Names the assumptions it made, so you can correct the ones that are wrong before deploying.
Getting better results
Where this usually goes wrong, and how to avoid it.
- Be specific in the limits field. 'Never give medical advice' produces a usable rule. 'Be careful' reads as tone and changes nothing.
- Keep the stop conditions. They are the most commonly deleted section and the reason agents loop or run up a bill overnight.
- Test the two things it names. The setup notes flag where the prompt is most likely to be wrong for your case, which is faster than discovering it in use.
More AI Agents prompts
All AI Agents →- AI Agents
Personal Assistant Agent System Prompt
Generates a personal assistant prompt with your preferences, your priorities, what it may decide alone, and what it must always ask you about first.
ClaudeChatGPTGemini006Open → - AI Agents
Research Agent System Prompt
Generates a research agent prompt that cites sources, states its confidence, separates fact from inference, and refuses to fill gaps with plausible invention.
ClaudeChatGPTGemini002Open → - AI Agents
Data Analysis Agent System Prompt
Generates an analysis agent prompt that states its assumptions, checks data quality before concluding, refuses causal claims the data cannot support, and shows its working.
ClaudeChatGPTGemini002Open →
Written for The AI University. Every prompt in this library is original work — authored, tested and revised here, not collected from elsewhere. 365 of them, free with an account.