These labs produce evidence for the IDEAL Lab IT & Research Computing Passport. Complete only the labs required by your selected tracks.
| Lab | Main evidence |
|---|---|
| First safe pull request | Focused PR with real verification |
| Manual versus agent | Comparable task, reviewed agent diff, and personal verification |
| Reproducible Python | Environment recreated from declarations |
| Data placement | Approved data-location map |
| Incident scenarios | Safe containment and escalation decisions |
| First Euler job | Tiny job plus accounting output |
| Euler resource optimization | Evidence-based right-sizing |
| Safe Euler arrays | Explicit justified concurrency cap |
| Euler GPU review | Safe one-GPU plan without required submission |
| Project handover | Reproducible transfer packet |
Lab Rule#
An AI agent may explain, review, or suggest a small patch. It must not claim the student ran a command, understood a concept, or obtained output that the student did not personally verify.
The first code change is completed manually. The agent comparison comes only after the learner can inspect a branch, diff, test, commit, and pull request without delegating those concepts.