Policies & Guards

The Policies & Guards panel defines and manages security policies that govern agent behavior across your fleet.

Key Features

• Policy definitions — security rules and guardrails that apply across all agents
• Guard configuration — automated enforcement of policy boundaries
• Violation tracking — log of policy violations with agent, timestamp, and details

How It Works

Policies work alongside the shield engine to provide governance-layer controls. While the shield focuses on prompt-level threat detection, policies address operational boundaries like which models agents can use, maximum token budgets, allowed actions, and time-of-day restrictions.

Guards are the enforcement mechanism for policies. When an agent action violates a defined policy, a guard can log, alert, or block the action depending on the policy configuration.