# Guardrail policy: definition and example

URL: https://aiagentsecuritysolutions.com/glossary/guardrail-policy/
Updated: 2026-11-16

A guardrail policy is a written rule that maps a detected condition to an action, such as alert, block, redact or log. It is separate from the detector that finds the condition. Teams tune the mapping per agent and per data class.

## Example

Cisco says its Project Quarry detection language is in production in AI Defense and maps findings to alert, block, redact, quarantine or log.

## Sources

- [Project Quarry: agent security detections](https://blogs.cisco.com/ai/project-quarry-agent-security-detections) (blogs.cisco.com), read 2026-11-16

## Related pages

- [Agent observability](/glossary/agent-observability/)
- [LLM gateway](/glossary/llm-gateway/)
- [Autonomy level](/glossary/autonomy-level/)
- [AI Agent Security Best Practices, Layer by Layer](/guides/ai-agent-security-best-practices/)
- [AI Agent Security Testing and Red Team Exercises](/guides/ai-agent-security-testing-and-red-team-exercises/)
- [All glossary terms](/glossary/)
- [Goal hijacking](/glossary/goal-hijacking/)
- [Just-in-time access](/glossary/just-in-time-access/)
- [Guides](/guides/)
