> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.aurasos.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.aurasos.com/_mcp/server.

# Integration testing

Use staging to prove both the API contract and your operational behaviour before
requesting production access.

## Test setup

1. Use staging credentials and resolve staging response and subscription type
   IDs.
2. Register test customers with recognisable external references.
3. Configure a staging webhook receiver over HTTPS.
4. Use the **Responder Simulator** in the staging AURA portal to progress
   callouts through dispatch, arrival, completion, and closure.

> **Warning**
>
> Staging callouts are not closed by a live control room. The Responder
> Simulator is how you progress and close them, and each step emits the same
> webhook event as a live response. See
> [Testing with the Responder Simulator](/overview/core-concepts/webhooks#testing-with-the-responder-simulator)
> for the step-to-event mapping.

## Core scenarios

| Scenario          | What to verify                                                                    |
| ----------------- | --------------------------------------------------------------------------------- |
| Authentication    | Tokens are cached safely, renewed before expiry, and never exposed to the client  |
| Customer creation | Your external reference maps reliably to the returned `customerId`                |
| Subscription      | The customer is entitled only during the intended validity window                 |
| Callout creation  | Correct response type and location are sent; the returned `calloutId` is retained |
| Existing callout  | A repeated customer panic does not create an unintended duplicate experience      |
| Live tracking     | Location updates stop after closure; unknown statuses remain non-terminal         |
| Cancellation      | Customer and trusted-backend cancellation paths behave as designed                |
| Webhooks          | Signalling is handled asynchronously and duplicate events are safe                |
| Rate limits       | `429` and `RateLimit-Reset` produce backoff rather than a retry storm             |
| Transient failure | Timeouts and `5xx` do not create unsafe duplicate writes                          |

## Product-specific scenarios

For **People Protect**, test poor GPS accuracy, background/foreground movement,
an expired subscription, and customer cancellation.

For **Property Protect**, test an unknown external reference, a deleted Fixed
Location, CSV validation and callbacks, and an alarm for a site that already has
an active callout.

For **white-label**, test first launch, authentication, subscription choices,
panic and tracking, support links, notifications, deep links, and store-ready
builds for every enabled brand and environment.

## Evidence for go-live

Keep a short test record containing the scenario, timestamp, environment,
request correlation or resource IDs, observed result, and owner. Never paste
tokens, secrets, full customer profiles, or payment data into the record.

Continue with [Production readiness](/overview/deploy/production-readiness) and
the go-live checklist for your product.