Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Saurabh-07586/examplatform/llms.txt

Use this file to discover all available pages before exploring further.

This reference covers every proctoring constraint, monitored signal, and risk score level used by ExamPlatform’s AI proctoring engine. During an active exam session, the platform continuously analyses the student’s webcam feed, browser state, and network connection to generate a real-time integrity score. Admins and examiners can review this data in the Live Monitoring dashboard and in post-exam AI Proctoring Reports.

Browser Requirements and Restrictions

Before a student can begin an exam, the following browser conditions must be met and are enforced throughout the session.
  • Fullscreen mode is mandatory. The exam interface enters fullscreen automatically at launch. Exiting fullscreen at any point is logged as a fullscreen_exits event and may raise the student’s risk score.
  • Copy, paste, right-click, and printing are disabled. These browser actions are intercepted and blocked during the entire exam session to prevent content extraction or external assistance.
  • Camera and microphone access is required. Students must grant permission to both devices before the exam starts. If either permission is blocked or denied, the exam cannot proceed. Students will see a warning: “Camera or microphone access appears blocked. Please allow access in your browser settings, then reload this page.”
  • A stable internet connection is strongly recommended. The timer is server-controlled and cannot be paused. The exam auto-submits when time expires. Network drops are logged as connectivity_issues events.
  • A sufficiently charged device is recommended. A device losing power mid-exam will terminate the session; the auto-submit mechanism requires an active connection to save the student’s responses.
All browser restrictions are enforced at the application layer. Students using unsupported or heavily customised browsers may encounter compatibility issues. ExamPlatform officially supports the latest two major versions of Chrome, Edge, and Firefox.

Monitored Signals

The AI proctoring engine logs the following signals throughout the exam session. Each event is timestamped and stored in the session’s proctoring timeline, which is visible in post-exam reports.
SignalField NameDescription
Gaze Awaygaze_away_eventsDetected when the student’s eye gaze deviates from the screen for a sustained period, indicating possible reference to external material.
Face Not Detectedface_not_detectedThe webcam feed no longer contains a recognisable face — the student may have moved out of frame or covered the camera.
Multiple Facesmultiple_facesMore than one face is detected in the webcam frame simultaneously, suggesting the presence of another person.
Mobile Device Detectedmobile_detectionsA secondary mobile device is visible in the webcam frame, which may indicate the student is using it to reference answers.
Tab Switchestab_switchesThe student navigated away from the exam tab. The total count is logged and contributes to the risk score calculation.
Fullscreen Exitsfullscreen_exitsThe student exited fullscreen mode. Each exit is logged individually with a timestamp.
Connectivity Issuesconnectivity_issuesA network interruption was detected. The platform records the start and duration of each connectivity gap.
gaze_away_events
integer
A running count of gaze-away detections for this session. Increments each time the AI determines the student’s eyes are directed away from the screen for longer than the detection threshold.
face_not_detected
integer
A running count of frames or intervals in which no face was detected in the webcam feed. Consecutive detections are grouped into a single event in the proctoring timeline.
multiple_faces
integer
A running count of intervals in which two or more faces were simultaneously visible in the webcam frame.
mobile_detections
integer
A running count of frames in which a mobile device was identified in the webcam feed by the object detection model.
tab_switches
integer
Total number of times the student left the exam browser tab during the session. Counted using the browser’s Page Visibility API.
fullscreen_exits
integer
Total number of times the student exited fullscreen mode during the session.
connectivity_issues
integer
Total number of distinct network interruption events detected during the session.

AI Risk Score

At the end of each exam session, the AI proctoring engine aggregates all monitored signals into a single risk level. This score is displayed in the Live Monitoring panel during the exam and in the AI Proctoring Report afterwards.
Risk LevelDescription
NormalNo significant anomalies detected. Monitored signals are within expected ranges for the duration of the session.
ElevatedSome suspicious signals were recorded — for example, three or more tab switches, or repeated gaze-away events. The session warrants a review but does not constitute confirmed misconduct.
CriticalStrong integrity signals detected that require immediate human review. Examples include multiple-face detections, persistent face-not-detected intervals, or a combination of several elevated signals across the session.
Risk levels are calculated automatically after the session ends and are updated in real time during the exam for live monitoring purposes. The thresholds used to transition between levels are configured at the platform level by your ExamPlatform administrator.

Viewing Proctoring Data

During the Exam — Student Monitoring Panel

While an exam is in progress, each student can see their own AI Monitoring panel on the exam interface. This panel displays the following indicators:
IndicatorDescription
WebcamConfirms the camera is active and the student is in frame.
MicrophoneConfirms the microphone is active and capturing audio.
Eye TrackingIndicates whether gaze is currently detected as focused on the screen.
Face DetectionShows whether a face is detected in the current webcam frame (displays “Face missing” when no face is detected).
Mobile DetectionFlags if a mobile device is visible in the webcam feed.
Tab SwitchesRunning count of times the student has left the exam tab.
AI Risk ScoreUpdates in real time as proctoring events are recorded.
InternetShows connection quality: Stable, Unstable, or Poor.
BatteryAlerts the student if their device battery is low.

During the Exam — Live Monitoring (Admin)

Admins and invigilators can access the Live Monitoring view from the exam management dashboard. For each active session, they can see the per-session current risk score, cumulative tab switch count, current question, time remaining, and the timestamp of the last successful server heartbeat. Admins can also trigger a Force Submit action to immediately end and submit a student’s session from the live monitoring panel.

After the Exam — AI Proctoring Reports

Once an exam session is complete, a full AI Proctoring Report is generated for each student. The report includes:
FieldDescription
Overall exam integrity statusThe final integrity verdict for the session.
AI Risk ScoreThe final aggregated risk level: Normal, Elevated, or Critical.
Suspicious activity timelineA complete timestamped log of all flagged proctoring events during the session.
The report also includes a breakdown of event counts by signal type — gaze away events, face not detected, multiple faces, mobile detections, tab switches, fullscreen exits, and connectivity issues — as well as snapshots or clips associated with flagged events (where supported by your plan).
AI proctoring is advisory. The risk score and flagged events are tools to help examiners and administrators identify sessions that may warrant closer review — they are not a determination of misconduct. All final academic integrity decisions must be made by a qualified examiner or administrator after reviewing the full proctoring report in context.

Build docs developers (and LLMs) love