Skip to content

Simulate a policy decision

Test policy intent before changing runtime behavior.

Have the proposed policy version and representative principal, action, resource, and context values.

  1. Open the policy simulation flow for the target environment.

  2. Enter an intended allowed request using the same identity and resource relationships as the real client.

  3. Repeat with a forbidden tool, missing permission, and an explicit deny that overlaps an allow.

  4. Read the decision and reason for each case. Correct the draft until the cases match the intended access boundary.

Save a metadata-only list of tested cases and expected decisions. After publication, verify the real client produces the same outcomes.

Simulation does not prove route health, credential availability, or replica activation. Diagnose those independently when the deployed call fails.

Read the related concept or consult the Admin API for exact request fields.

Read the source contract.

Amelfi documentation · Space Grotesk / Geist Mono