Agent rules should be enforced outside the prompt
Rules written only inside a prompt are not reliable controls for an . The model may ignore them, earlier context can lose influence as the session gets longer, and the model may behave as if the rule no longer exists. Real rules should be enforced by the around the model.
The should decide which tools the agent may run, which data it may read, and which actions it may take. Without that outside control, a rule in the prompt is closer to a request than a hard limit.
Key points
- A prompt rule is not the same as an enforced system limit.
- can weaken the effect of earlier context.
- Tool access, data access, and allowed actions should be controlled by the .
- Relying on longer prompts can raise token use without solving safety problems.