Skip to main content

Your First Policy

A policy in Actra defines when an action should be allowed or blocked before execution.

Let’s break down a simple rule.

Example Policy​

version: 1

rules:
  - id: block_large_refund
    scope:
      action: refund
    when:
      subject:
        domain: action
        field: amount
      operator: greater_than
      value:
        literal: 1000
    effect: block

Rule Anatomy​

This rule reads as:

Block the refund action when amount > 1000

Every rule is made of 5 important parts.

1) Rule ID​

id: block_large_refund

A unique identifier for the rule.

This is returned in policy errors and debugging output.

2) Scope​

scope:
  action: refund

Scope limits the rule to a specific action.

This rule applies only when the protected function is admitted as:

runtime.admit("refund", fn)

3) Condition​

when:
  subject:
    domain: action
    field: amount

This tells Actra:

inspect the amount field from the action input.

4) Operator + Value​

operator: greater_than
value:
  literal: 1000

This compares:

action.amount > 1000

5) Effect​

effect: block

If the condition matches, the function execution is blocked.

Mental Model​

Think of a rule as:

For this action,
if this condition is true,
apply this effect.

That simple model scales to:

  • workflow approvals
  • AI tool safety
  • deployment controls
  • runtime governance

Next Steps​

  • Learn the full Schema DSL
  • Explore Actor and Snapshot
  • Write multi-rule policies