Sprint 4 final user testing¶
Team-approved final scope (9 October 2026): two genuine human sessions, both recorded. One public/analyst session and one assisted local submitter/reviewer session cover complementary workflows. The AI browser simulation is supplemental and excluded from the human count; partial outcomes and evidence limitations remain disclosed.
Sprint 4 reuses the protocol, task bank and ADR-013. Automated checks and facilitator walkthroughs prepare the environment; they do not count as representative-user sessions.
| Planned session | Exact selected Task IDs | Human evidence |
|---|---|---|
| Public / analyst | PUB-01; PUB-02; PUB-03; PUB-06 | P15 reviewed; four Success outcomes; decisions evaluated |
| Approved submitter | AUTH-01; AUTH-02; SUB-01; SUB-02; SUB-03; SUB-04 | Assisted local session; Partial outcomes; metadata/feedback pending |
| Reviewer | AUTH-01; REV-01; REV-02; REV-04 | Same local session spans both workflows; one combined human session |
| API consumer (recommended) | PUB-05; API-01; API-02 | Pending |
Use the version-controlled Sprint 4 evidence pack, including the participant/facilitator guides, actual session records, findings, decisions, retests and limitations.
The team unanimously approved a final scope of two genuine human sessions on 9 October because of the submission deadline, rather than treating its original three-session plan as a course-imposed minimum. Both real sessions are documented. Preserve the actual build/environment, recorded task scores, assistance, findings, decisions, retest limitations and outstanding privacy/participant-information limitations; do not treat the simulation or the two workflows in the combined session as another participant. Sprint 2/3 evidence remains separate.
810 remains the final release gate; this preparation page is not release acceptance.¶
The 7 October completion audit records the earlier three-session plan and observed limitations. Its 9 October decision addendum documents the unanimous two-session scope approval without rewriting the historical 7 October findings. Both local fixtures published with 4 runs, 1 wicket and 2 legal balls; the deliberately invalid package remained rejected. Navigation and corrected input were supplied during the local exercise, so the nine outcomes remain Partial.
AI Declaration¶
Original preparation: Codex[GPT-6], with P15 outcomes and decisions reviewed by Gabriel. The 9 October team-approved scope reconciliation was drafted with ChatGPT-Web[GPT-6]; approval was reported by the student auditor. Missing participant/technical evidence is not treated as observed.