Enumeration GuardrailAction

High-level outcome emitted by a guardrail evaluation.

The action instructs AgentOS how to handle evaluated content:

• GuardrailAction.ALLOW - Pass through unchanged
• GuardrailAction.FLAG - Pass through but record metadata
• GuardrailAction.SANITIZE - Replace content with modified version
• GuardrailAction.BLOCK - Reject/terminate the interaction

Example

// Allow content to pass
return { action: GuardrailAction.ALLOW };

// Block harmful content
return {
  action: GuardrailAction.BLOCK,
  reason: 'Content violates policy',
  reasonCode: 'POLICY_VIOLATION'
};

// Redact sensitive information
return {
  action: GuardrailAction.SANITIZE,
  modifiedText: text.replace(/\b\d{3}-\d{2}-\d{4}\b/g, '[SSN REDACTED]')
};
Copy