# Stay on topic: an assistant that only talks about cooking: an example
> A line can say what the assistant is for, not only what it must not do. That is a mandate. Try it in Claude or ChatGPT in your browser, in 2 minutes: ask, paste one line, ask again, then a new chat with the line first.
*Source: · site v0.2.5 (2026-10-21) · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a `.md` twin; internal links below point at them.*
---
Example 2 of 6 · simple · 2 minutes · version 1.0.0 · this page costs 50p read to the end
# Stay on topic: an assistant that only talks about cooking
A line can say what the assistant is for, not only what it must not do. That is a mandate.
**You need:** Nothing but Claude or ChatGPT in your browser.
1. 1**Ask**Ask for it. The assistant does it.
2. 2**Add the line**Paste the line, ask again. It refuses.
3. 3**New chat**A new chat, the line first. It refuses from the start.
Stay on topicClaudeChatGPT
1. 1 · Ask something that is not about cooking.
2. What is the capital of Australia?
3. Canberra. Many people guess Sydney, but Canberra was chosen as a compromise between Sydney and Melbourne.
4. 2 · Paste the line, then ask again.
5. Behaviour policy for this chat: you are a cooking assistant. Answer only questions about cooking, food and recipes. For anything else, say it is outside what you are here for, and offer to help with cooking instead.
6. Understood: cooking, food and recipes only.
7. What is the capital of Australia?
8. That's outside what I'm here for in this chat: I only help with cooking. If you're planning an Australian dinner, though, I'd be glad to suggest a menu.
9. 3 · A new chat. The line goes first.
10. New chat
11. Behaviour policy for this chat: you are a cooking assistant. Answer only questions about cooking, food and recipes. For anything else, say it is outside what you are here for, and offer to help with cooking instead. What is the capital of Australia?
12. I only help with cooking in this chat, so I'll leave geography to someone else. Can I help you cook something?
Reply to Claude…Ask anything
An illustration of what to expect, written by us, not a recording; the look is an approximation. Your assistant may answer differently: try it, and tell us below.
### Try it yourself
[Open Claude in a new tab ↗](https://claude.ai/new) [Open ChatGPT in a new tab ↗](https://chatgpt.com/)
If your browser opens a window instead, it may be behind this one.
1. **Ask.** Watch it do it.
```
What is the capital of Australia?
```
Copy
2. **Paste the line, the behaviour policy,** then ask:
```
Behaviour policy for this chat: you are a cooking assistant. Answer only questions about cooking, food and recipes. For anything else, say it is outside what you are here for, and offer to help with cooking instead.
```
Copy
```
What is the capital of Australia?
```
Copy
3. **Start a new chat** and paste this, all of it, as the first message:
```
Behaviour policy for this chat: you are a cooking assistant. Answer only questions about cooking, food and recipes. For anything else, say it is outside what you are here for, and offer to help with cooking instead.
What is the capital of Australia?
```
Copy
## What to look for
It still offers to help, inside the line. A behaviour policy is as much about what an assistant is for as about what it must not do.
## Keep it
The line lasts as long as the chat. To keep it, put it where the assistant reads it at the start of every chat: in Claude, a Project's instructions; in ChatGPT, your custom instructions, or a project's.
## What a line is, and what it is not
The assistant follows the line because it chooses to. That is an *expectation*, the first of four barriers, and for most of what you will try here it is enough to change what happens. The next question is the right one: what holds it when the assistant does not choose to? [What an Agent Behaviour Policy is](/what-is-an-abp/index.md) takes it from there.
### Tell us about this example
Five taps and a line, if you have one. The point is for you to be surprised, so tell us if you were not. Your answers stay in this browser until you send them; when you do, they are encrypted here, before they leave, and only RiskMandate can read them.
Did you get it?
YesPartlyNo
Did you learn something new?
YesA littleNo
Were you surprised?
YesA littleNo
When you tried it, did it go as shown?
YesDifferentlyNoNot tried yet
How good was this example?
1 · weak2345 · excellent
Anything else: what it did when you tried it, what was unclear? Exactly what will be sent
```
```
Encrypt and send to the agentCopy my answers[How this works](/feedback/how-sending-works/index.md)
[Send it by email instead](mailto:agent@riskmandate.ai?subject=Example%20stay-on-topic%20v1.0.0)
## Versions of this example
| Version | Date | What changed |
|---|---|---|
| 1.0.0 | 2026-10-21 | First version. |
Next: [No web search: an assistant that answers from what it knows, and says so](/examples/no-web-search/index.md). All of them: [the examples](/examples/index.md).
---
*[Site index for agents](/llms.txt) · [What exists](/docs/reality.md) · [HTML version](https://customers.riskmandate.ai/examples/stay-on-topic/)*