Topic Confinement
Restrict the agent's conversational and functional scope to its declared domain — a banking agent refuses to write code; a code assistant refuses to give medical advice.
This page is the complete instruction page for one pattern called "Topic Confinement." It explains the problem this pattern solves, the idea behind the solution, when you should (and should not) use it, and what happens afterward — both the good effects and the costs.
This matters because building AI agents is not just about making them clever. It is also about making them safe and predictable. Following a well-tested pattern like this one helps avoid common mistakes, and shows you exactly which safety rules and regulations it connects to, listed under "Standards Mesh" on this page.
Context
Generalist models can do everything; specialist agents should not. Out-of-scope engagement wastes resources, confuses users, and creates liability.
Problem
Unconfined agents drift into topics outside their governance envelope.
Forces
- User experience of refusal vs scope discipline
- Brittleness vs clarity
Solution
At input, classify whether the request is in-scope. If not, respond with a scope-aware refusal that explains what the agent can help with and where to go for the off-topic need. Classifier is cheap; refusal pattern is standardised.
Applicability
- Specialist agents
- Regulated-sector agents
- Customer-facing agents with clear domains
Anti-Patterns
- Agent that answers everything because the user insists
- Classification entirely inside the main prompt
Consequences
- +Clear scope
- +Reduced regulatory exposure
- −Occasionally unhelpful on edge cases
- −Classifier requires maintenance