AG
Interactive Rubric BuilderDEMO
CS704 Applied Agentic AIRubric v2.1

Interactive Academic Rubric Builder

Map technical test assertions, behavioral sandboxing, and oral viva defenses onto academic learning outcomes.

Total: 100 / 100 marks(Balanced)
Published Versionv2.1
StatePublished
Last PublishedMar 1, 2027
Published ByDr. Sarah Bennett
Official Course Assessment Standard for CS704 Assignment 2

Assignment 2 Criteria Configuration10 Criteria

Drag or use Up/Down arrows to adjust hierarchy. Calibrate weight distribution to maintain 100 total marks.

#1

Task Completion

hybridgeneralViva (10%)

Autonomous research goals achieved with full analytical synthesis and report generation.

Evaluation Balance: 70% Automated Test Assertions • 30% Lecturer Moderation

20 marks
#2

Evidence Grounding

hybridgeneral

All synthesized statements strictly attributable to retrieved corporate filings or reputable web data.

Evaluation Balance: 80% Automated Test Assertions • 20% Lecturer Moderation

15 marks
#3

Tool Selection & Dispatch

hybridgeneral

Efficient, logical sequencing of web_search, doc_retrieval, and python_repl without redundant calls.

Evaluation Balance: 60% Automated Test Assertions • 40% Lecturer Moderation

15 marks
#4

Citation Integrity

automated onlygeneral

Precise attribution, DOI/URL validity, and avoidance of hallucinated bibliographies.

Evaluation Balance: 100% Automated Test Assertions • 0% Lecturer Moderation

10 marks
#5

Repeatability & Consistency

automated onlygeneral

Consistent execution success and semantic agreement across 5 stochastic runs.

Evaluation Balance: 100% Automated Test Assertions • 0% Lecturer Moderation

10 marks
#6

Prompt Injection Resistance

hybridgeneralViva (20%)

Resilience against direct and indirect adversarial document injections (e.g. PI-017 test suite).

Evaluation Balance: 60% Automated Test Assertions • 40% Lecturer Moderation

10 marks
#7

Error Recovery & Reflection

hybridgeneral

Graceful handling of tool timeouts, empty search payloads, and malformed JSON responses.

Evaluation Balance: 50% Automated Test Assertions • 50% Lecturer Moderation

5 marks
#8

Efficiency & Resource Budget

automated onlygeneral

Execution under 5.0s latency, less than 20,000 tokens, and under $0.05 estimated cost per report.

Evaluation Balance: 100% Automated Test Assertions • 0% Lecturer Moderation

5 marks
#9

Code Quality & Modularity

manual onlygeneral

Clean separation of agentic memory, prompt templates, tool definitions, and exception handlers.

Evaluation Balance: 0% Automated Test Assertions • 100% Lecturer Moderation

5 marks
#10

Architecture Documentation

manual onlygeneral

Comprehensive README detailing agent loop, state graph, threat assumptions, and setup.

Evaluation Balance: 0% Automated Test Assertions • 100% Lecturer Moderation

5 marks

Published Revision History

Immutable Audit Trail
v2.1Active• Published by Dr. Sarah Bennett

Calibrated indirect prompt injection resistance test threshold (PI-017) and 100-mark balance.

Mar 1, 2027
v2.0• Published by Dr. Sarah Bennett

Initial release of CS704 Assignment 2 Master Assessment Rubric baseline specification.

Sep 1, 2026