Back to submissionsProvisional result
Lecturer review required
Tom Cruise / Agentic Student Support Workflow
5 of 16 tests passed / Evidence completeness 90%
AI-generated evaluation is advisory. The lecturer reviews the evidence and makes the final academic decision.
31%
Provisional overall mark
Borderline evidence profile
5 tests passed 11 tests failedRubric breakdown
Functional correctness
38%35% weighting / 13.3 marks
Agentic workflow and orchestration
32%20% weighting / 6.4 marks
Safety and human oversight
20%15% weighting / 3.0 marks
Testing and reliability
22%15% weighting / 3.3 marks
Code quality
36%10% weighting / 3.6 marks
Documentation
25%5% weighting / 1.3 marks
Evidence and observations
Strengths
- The project compiles and exposes the expected entry point.
- A structured object is returned for the sample message.
Areas for improvement
- Responses are hard-coded and do not reflect the message.
- High-risk messages are not escalated.
- The extension response states that the request has been approved.
- No automated student tests were included.
Code-level observations
- src/index.ts returns the same category and response for every input.
- requiresHumanReview is always false.
- tests/ contains no student-authored scenarios.
Safety evidence
- The extension draft claims approval, contrary to the brief.
- High-risk input does not set requiresHumanReview.
Personalised feedback
Editable draftRecommended learning actions
- Replace the fixed response with separate classification and risk steps.
- Ensure High risk always sets requiresHumanReview, and remove approval claims.
- Add tests for empty input, high-risk wellbeing, and extension wording.