# App interruption test worksheet

Use a disposable app and fictional records. Every observation below starts at **Not tested**. A passing screen check does not establish private storage, payment completion or external delivery.

Related guide: https://www.overskill.com/learn/test-ai-app-interruptions

## Define one journey

- App and version:
- Tester and date:
- Device/browser:
- Intended task:
- Starting page:
- Final outcome the person should receive:
- Current draft-saving promise:
- Submitted-request expiry policy:
- What must require explicit confirmation:
- External actions excluded from this pass:

Fictional example: Maya Chen requests a consultation for **September 17** with the message **“Discuss the sample workshop schedule.”** Use a test account you control; the name is not an existing account. Use Luis Rivera as a second, separately controlled account. Record the selected service and timezone if relevant.

Keep the distinctive text unchanged during each journey so you can spot lost fields. Create a fresh request for checks that accept, cancel or expire one; identify each with a safe local label. A previously cancelled request cannot also establish expiry behavior. Redact account addresses, private identifiers and real customer data from shared evidence.

## Run and record

Status choices: **Not tested**, **Pass**, **Fail**, or **Not applicable** with a reason. Use Pass only after checking the actual expected outcome.

| Check | Expected | Observed | Status |
| --- | --- | --- | --- |
| 1. Enter a draft, reload, then leave and return | Behavior matches the visible saving promise. No other account's draft appears. | Not tested | Not tested |
| 2. Begin signed out, submit and follow actual sign-in | The correct service, date and message remain; the person can find the task again. | Not tested | Not tested |
| 3. Inspect the review before confirming | Current account, destination and details are clear. The confirmation names its consequence. | Not tested | Not tested |
| 4. Switch to the second account and reopen the old page | Private details remain unavailable and the other account cannot continue the request. Test workspace switches separately if supported. | Not tested | Not tested |
| 5. Repeat an accepted submission in a safe test environment | The same result is identified; no duplicate record or action is created. Inspect actual records. | Not tested | Not tested |
| 6. Cancel a pending request, then use Back or its old link | It remains cancelled; a stale page cannot restart it. | Not tested | Not tested |
| 7. Return after documented expiry | The old request stays unavailable, with an explicit path to begin again. Its lifetime is not silently renewed. | Not tested | Not tested |
| 8. Inspect an accepted request whose later processing fails | Status identifies the existing result and explains what a recovery action repeats. No misleading completion message. | Not tested | Not tested |

## Keep a result receipt

For each performed check, record:

- Check number:
- Exact action and input:
- Account/workspace role used:
- Expected outcome:
- Observed outcome:
- Request/result reference, if safe to retain:
- Record count before/after a retry:
- Screenshot or other evidence location:
- What was not exercised:
- Repair requested:
- Repeat-check result and date: **Not tested**

Separate **request received**, **result created** and **later processing completed**. For payments, messages or other external actions, use a supported safe test mode where available and establish what a retry does before trying it. If no safe setup is available, leave that execution Not tested.

## Retest the complete journey

- Original successful journey after the repair: **Not tested**
- Phone-sized journey: **Not tested**
- Keyboard journey: **Not tested**
- Second account remains unable to read or change private work: **Not tested**
- Cancelled/expired requests remain unavailable: **Not tested**
- Actual external outcomes, if in scope: **Not tested**
- Temporary records and account state restored:

Release decision:

- Essential checks still failing or Not tested:
- Owner of each remaining check:
- What users will be told about saving, expiry and recovery:
- Evidence supporting readiness for the intended use:

For broader testing, use https://www.overskill.com/learn/ai-app-launch-checklist and https://www.overskill.com/learn/compare-ai-app-builders.