Common C02 Stage 1 testing checklist

c02-stage1-public-review-2026-10-23.2 · Review start page

Owner: physical iPhone, iPad and Mac touch, scrolling, orientation and usability. ChatGPT: independent browser interactions where available. AL: functional/evidence and keyboard/accessibility review. Use the same checklist; report coverage rather than assuming each reviewer tested every device.

Before testing, record build ID, full URL, requirement, device/OS/browser version, dimensions/zoom and input method. Main route: answer_and_method. Use a fresh page for each independent scenario; refresh deliberately clears the fictional session. Do not refresh halfway through a preservation test.

Use Pass, Fail, Not tested or Blocked per scenario. Never infer a pass from absence of observation. Physical-device, simulated viewport, real lifecycle and screen-reader tests are separate evidence categories.

  1. C01 — Exact unfinished working

    Enter 012 in Top number, × and 004 in working. Pause; verify controls disappear. Resume: values and Working disclosure remain exactly as entered. Also try blank and partial input. Do not treat Pause as Check.

  2. C02 — Original answer and correction

    On a fresh page, check 9. Edit to 012 without checking. Pause/resume. The draft remains 012; the inspector still records the original checked 9, with no new assessment.

  3. C03 — Answer versus method

    Check 12 with ÷4. Test each requirement in a separate page. Numerator correctness is recognised, method remains unresolved; only answer-only may satisfy its answer requirement. Pause/resume preserves feedback and adds no conclusion.

  4. C04 — Selected versus displayed groups

    On a blank page choose Help. Show three groups (15), select four without Show groups, pause/resume. Four stays selected but the displayed model remains three/15 until you press Show groups.

  5. C05 — Unfinished counting

    On a fresh page choose ×4, leave the numerator blank, then Help. Enter 01 in Total number of counters without Try total. Pause/resume: 01 stays, focus returns appropriately, and the fraction is not filled automatically.

  6. C06 — Continue the same Help

    Show groups 3 then 4, pause/resume, then genuinely choose another count. Choices remain linked to the same support request; each total is labelled as a system consequence.

  7. C07 — Presentation timing

    Try pausing immediately as Hint/Help opens, then resume. Watch for duplicate request/presentation records or an unusable support screen. This manual check is not a deterministic inter-animation-frame test. The controlled timing harness is intentionally excluded from the public package.

  8. C08 — Repeated Hint/Help pauses

    Open Hint and then, in a separate check, Help. Repeat Pause/Resume several times. The same screen, draft and recorded presentation remain; no duplicate presentation just from Resume.

  9. C09 — Keyboard/pointer isolation

    During Pause, Tab through visible actions. Hidden answer/support controls must not receive focus or activation. Enter on Resume must not Check an answer. Confirm a visible focus indicator. Do not activate Finish or restart unless that is the test.

  10. C10 — Explicit Finish

    While paused, choose Finish for now. This should add only the explicit conclusion, not a fabricated answer, success, mastery or engine pause/resume record. Resume should no longer be offered.

  11. C11 — Restart

    Pause with unfinished Help input. Start a new test, then Keep this session: pause and working must survive. Separately confirm Start new session: a clear warning precedes the reset, and the new fictional page is blank.

  12. C12 — Session limits

    If a normal interaction reaches a demonstration limit, Pause/Resume must not clear it; Finish/restart remain explicit. Do not interpret the technical limit as failure. Automated limit coverage already exists; record Not tested if you do not exhaust it manually.

  13. C13 — Rapid cycles

    Pause/resume repeatedly, including while support is opening. No lost inputs, unexpected checks, duplicate presentation or focus trapped in hidden content.

  14. C14 — Evidence/report integrity

    Compare the inspector before/after Pause/Resume. No engine pause/resume records or parent/tutor pause claims. Existing responses/support sources remain unchanged. Editing is not a submitted correction; displayed totals are not independently calculated pupil answers.

  15. C15 — Device/layout/accessibility

    Try portrait/landscape, keyboard open, narrow window, dark and large-text routes. Controls and feedback remain reachable through ordinary scrolling, without hidden focus or horizontal clipping. Separately record actual screen-reader pairing and announcements if tested.

  16. C16 — Real lifecycle, separate from simulation

    While paused, background the browser/tab and return without refresh; resume and check values. Separately try backgrounding as support opens and OS sleep/return. Record real device/browser and whether memory survived. If the browser reloads/discards, record that: recovery is excluded, not a promised save. Do not label a simulated visibility test as a real lifecycle pass.

Evidence and sharing

Record exact steps and fictional values. Inspect the learning inspector's source-linked record if possible. Do not enter or share real pupil data. Reports are not uploaded or shared automatically; copy only the relevant fictional observations into your review response.

Download/copy result template

Reviewer/role:
Build ID and manifest checked:
URL including requirement/appearance:
Device, OS and browser/version:
Viewport, zoom, orientation, input method:
Screen reader/version (or not used):
Scenario ID:
Result: Pass / Fail / Not tested / Blocked
Exact starting values and steps:
Expected / actual:
Evidence record IDs or screenshot (fictional only):
Real background/device test OR simulated:
Reproducible? Frequency:
Notes / remaining limits:

Known limits

Same running page/tab only. No permanent save, refresh recovery, discarded-page recovery or cross-device continuation. No fresh-question transfer. No new responsive architecture. No claim of mastery from one answer. Actual screen-reader announcements, physical devices, real background/OS behavior and pupil usability remain acceptance work. The exact pause-between-two-animation-frames interleaving also remains a focused technical check.