Templates / "Ask before you act" rules for any agent

"Ask before you act" rules for any agent

Paste into the instructions of ChatGPT, Claude, Grok, Zapier, n8n or a custom GPT so it checks with you before anything that can't be undone.

StarterOpen, no sign-up

Who it is for

Anyone giving an AI assistant or agent the ability to send, post, delete, buy or book.

How to use it

Paste it into the instructions of your assistant or agent.
Test it by asking for something that sends or posts, and check that it stops and asks.
For more specific rule sets (email, calendar, shopping, accounts), see the variants template.

The template

Before you send, post, delete, buy, book, or change anything outside this chat:
1. Tell me in one line exactly what you're about to do and to whom.
2. Show me the final text, amount, or change.
3. Wait for me to reply "yes" before doing it.
Reading, searching, summarizing and drafting are fine without asking.
If you're unsure whether something counts, ask.

Guardrails and human approval

A prompt is not a lock. Also limit the assistant's real permissions.
Review what it did regularly.

These are practical notes, not legal advice. Templates are provided as is. See section 4 of the Terms.

Works with

ChatGPT, Claude, Grok: Paste it as your message, then add your own details where the [BRACKETS] are.
Custom GPTs, Claude Projects, Grok or other assistant projects: Paste it into the instructions or project instructions field if you want to reuse it. Check that your plan includes that feature.
n8n, Zapier, Make and other automation tools: Use it as the system message or prompt of an AI step. Keep a person approving anything that gets sent or posted.

We have not benchmarked output quality across models. Output varies by model and by what you paste in, so read it before you use it.

Related templates

'Ask before you act' variants for personal assistants Research agent guardrails All templates