> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rotascale.com/llms.txt
> Use this file to discover all available pages before exploring further.

# The six outcomes

> allow, deny, exhausted, gated, review_sync and review_async are different findings with different remedies, which is why they are not a boolean.

An authorisation returns one of six outcomes. They are not collapsed into
allowed/refused because the four refusal shapes send an operator to completely
different places.

| Outcome        | Allowed | What it means                                                    | What to do                                              |
| -------------- | ------- | ---------------------------------------------------------------- | ------------------------------------------------------- |
| `allow`        | yes     | Within the grant.                                                | Proceed.                                                |
| `deny`         | no      | Policy refused it.                                               | A policy conversation.                                  |
| `exhausted`    | no      | Permitted, but the allowance is spent.                           | Retrying cannot help. Escalate or wait for the window.  |
| `gated`        | no      | Untrusted content entered and this action needs a clean context. | Escalate; see [Clean context](/concepts/clean-context). |
| `review_sync`  | pending | A person must decide before the action.                          | Block and surface the review.                           |
| `review_async` | pending | A person must decide, but not on this path.                      | Park and continue.                                      |

<Note>
  `deny` and `exhausted` are the pair most often merged by other systems, and
  merging them costs you the diagnosis. "You were not allowed to do that" and
  "you had run out" have different remedies and different owners.
</Note>

Each maps to an exception in the SDK. See [Errors](/errors).