A queue, the evidence, a decision with a reason, and an appeal. This is the workflow an institution has to be able to staff and defend.
Every violation with its severity and confidence. Every piece of evidence in a viewer. The full session timeline, reviewer notes and the case history.
Accept or dismiss each violation individually, assign to yourself, escalate to a review manager, then record one decision: approve, warning, manual review, invalidate or disqualify — with a reason. Closed cases can be reopened.
Each artifact is fingerprinted with a SHA-256 at the moment of capture, and that fingerprint is shown to the reviewer. Evidence bytes are never exposed by a public link — they are streamed through the backend under permission checks.
Candidates can appeal a decision. Appeals are tracked open → under review → resolved, with their own decision record. Backlog, cases closed, open appeals and average review minutes are reported.
Reviewers and review managers can see proctoring evidence. They cannot see a candidate’s uploaded answer files — that is a grading permission, deliberately kept separate from an integrity one.
Cases open automatically when risk crosses the threshold, sorted highest first, filterable by status, decision, reviewer, candidate, assessment and date. A case can also be opened by hand — an invigilator report is an integrity case even with no AI behind it.
