Skip to main content

Confirmation Requirements

Confirmation requirements gate operation execution behind explicit acknowledgment by the person the agent acts for.

When Confirmation Is Required​

  • risk_level is high or critical
  • The operation produces irreversible side effects
  • Financial transactions exceed defined thresholds

Declaration​

<button
  axag-intent="order.cancel"
  axag-entity="order"
  axag-action-type="write"
  axag-risk-level="high"
  axag-idempotent="true"
  axag-confirmation-required="true"
  axag-confirmation-message="This will cancel the order and initiate a refund. This cannot be undone."
>Cancel Order</button>

Agent Runtime Behavior​

When axag-confirmation-required="true", agent runtimes MUST:

  1. Present the confirmation message to the user
  2. Wait for explicit confirmation
  3. Only proceed upon affirmative response
  4. Log the confirmation event

Server Enforcement​

Confirmation that only happens in the agent can be skipped, so the server MUST be able to tell that a person confirmed this call. In the SDK:

  • The page's safety enforcers show a confirmation dialog the agent can't dismiss, and exchange the person's answer for a single-use token bound to the intent, the parameters and the actor.
  • @web-axag/server refuses a call without a valid token with AXAG_CONFIRMATION_MISSING. A token can't be reused, or replayed with different parameters.

The dialog shows the action's description, its risk level and the parameters the agent supplied. axag-confirmation-message isn't carried into the manifest in 1.1, so write the consequence into axag-description as well.