1. Data Protection (PII)
Block agents from sharing PII outside the company.
rules:
- id: block-pii-sharing
when:
output.contains("pii")
then: deny
message: "Sharing PII is not allowed"
SuperAgentX Policy Engine enforces your rules before any AI action happens — so every request is safe, compliant, and auditable by design.
Every request is evaluated in real time against your policies before any AI action is executed.
Define rules in simple YAML — enforce them across every agent, workflow, and tool.
Block agents from sharing PII outside the company.
rules:
- id: block-pii-sharing
when:
output.contains("pii")
then: deny
message: "Sharing PII is not allowed"
Use the right model based on the task type.
rules:
- id: route-by-task
when: task.type == "complex"
then: use_model: "gpt-4o"
else_if: task.type == "fast"
then: use_model: "gemini-1.5"
Restrict high risk actions in production environment.
rules:
- id: block-delete-prod
description: Block destructive API calls in production
scope: tool
when:
env: production
action: delete
decision: deny
Only allow specific teams to use certain models.
rules:
- id: model-access
description: Restrict model access by team
scope: model
when:
team: finance
model: claude-3
decision: allow
Require human approval for high-impact actions — with full context and audit trails.