Guardian icon

Guardian

Policy-based safety gate for AI agents and workflows. Check actions against Guardian policies, enforce previously-evaluated intents, or gate AI agent tool calls. Returns Allowed (including OBSERVED advisory decisions), Denied, or Needs Approval.

Actions3

Check → Confirm Execution

AI-generated

Summary

Confirms execution of a previously evaluated Guardian intent by posting its intent run ID to Guardian. Optionally supplies a payload for integrity verification; when left empty, the operation uses payloadJson from the incoming item if available.

Inputs

  • Intent Run ID (required) — The intentRunId returned by the Evaluate step.
  • Payload (for Integrity Check) — Optional payload to verify against the intent if payload hashing was enabled. Leave empty to use the incoming item's payloadJson object; to override, provide a JSON object or a valid JSON string, such as {{ JSON.stringify($json.payloadJson) }}.

Output shape

Three output branches: Allowed, Denied, and Needs Approval; Confirm Execution routes results to Allowed or Denied.

On a successful confirmation, Allowed contains Guardian's response fields and _guardian metadata including decision: ALLOW, intentRunId, executed: true, and executedAt. If the intent was already executed, the operation fetches the intent to distinguish a claim held by this workflow execution (Allowed, marked alreadyClaimed) from a duplicate claimed elsewhere (Denied, marked duplicateBlocked). API errors fail the operation unless Continue On Fail is enabled, in which case errors are routed to Denied.

Links

Discussion