Securing the Sandbox: A Rootless Containerized Framework for Process-Oriented Monitoring in Computer Graphics Education

Synthesis: Addresses a critical tension in CS education: the widespread adoption of generative AI makes it impossible to distinguish authentic student effort from AI code synthesis by evaluating final submissions alone. The paper presents VISMATIC, a rootless containerized framework that pairs robust environment isolation with explicit user-interaction tracking at the API level.

Key innovation: VISMATIC captures macro-level behavioral metrics (keystroke patterns, command execution timing, copy-paste behavior) that successfully flag statistical anomalies indicative of automated or off-platform workflows — while preserving student anonymity. Unlike JupyterHub, which exposes the host system to arbitrary shell commands, VISMATIC's rootless containers prevent lateral movement and host compromise.

The framework provides a scalable blueprint for preserving pedagogical integrity in the AI era, enabling educators to enforce authentic coding discipline through process-oriented monitoring rather than final-output evaluation. Open source on GitHub for institutional adoption.

What this means for practice

• Instructors. Treat telemetry as a screening signal, never as proof: the authors note that API-level event logs capture interaction patterns but not intent, so corroborate any flag with an oral interview about the flagged code, a comparison of exam performance against platform activity, and targeted plagiarism detection on notebook cells.
• Administrators. Use the pilot's cost profile to justify a controlled trial rather than enterprise hardware — VISMATIC ran for 19 active students, logging 1,880 discrete API events over approximately 57 active student-hours on a single Raspberry Pi 5 node rated for a load of 10 to 20 students.
• Instructors. Shift some assessment weight from the final artifact to the development process by reviewing session persistence, temporal rhythm, and workload composition alongside the submission, and use those signals for Formative Assessment checkpoints rather than only for integrity policing.
• Software developers. Keep student workloads on managed infrastructure and treat the closed course library and read-only materials as load-bearing controls, since the architecture depends on students being unable to download notebooks for local, offline execution.
• Administrators. Pair deployment with a published data-protection rationale: the framework avoids passive keep-alive telemetry and keeps behavioral logs separate from official academic records, which is what makes integrity monitoring defensible to students and regulators.

Limitations

• The evidence base is a pilot cohort of 19 active students in a single computer graphics course, measured as 1,880 discrete API events over approximately 57 active student-hours — far too few cases to validate anomaly detection thresholds or estimate false-positive rates.
• The authors state plainly that API-level event logs capture interaction patterns but not intent; a student may leave an automated script running for reasons unrelated to dishonesty or work offline and later upload solutions, so no flag in this study was corroborated against actual misconduct.
• There is no control group or side-by-side comparison with an unmonitored JupyterHub deployment, so the security advantage is argued architecturally rather than measured; isolation results come from one single-tenant deployment observed between March and May.
• The authors do not claim absolute security and describe the work as a risk-management approach, and the stated deterrence effect — that measuring dedication discourages shortcuts — is an expectation rather than an outcome the study measured.

Citation

Arroyo, G., López, L., & Torres, J. C. (2026). Securing the Sandbox: A Rootless Containerized Framework for Process-Oriented Monitoring in Computer Graphics Education.