Academic integrity forensics
Skill alainlebret/claude-agents/higher-ed-teaching-agents/skills/academic-integrity-forensics
Modular skill and workflow ecosystems for AI coding assistants. Structured, multi-agent pipelines for specialized professional domains.
npx -y skills add alainlebret/claude-agents --skill academic-integrity-forensicsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Detects likely plagiarism or collusion patterns across submissions using multi-signal similarity analysis and produces evidence-first integrity reports for human review.
SKILL.md
1.5 KB, as published. Nobody here has run it
Academic Integrity Forensics
Goal
Identify suspicious similarity clusters across student submissions while preserving fairness, explainability, and teacher control.
Inputs
mission.jsonrubric.mdsubmissions/and/orevidence/- Optional logs or metadata (timestamps, commit traces, test traces)
Produce
Write:
integrity-report.json(global integrity assessment)integrity-summary.csv(per-student risk indicators)integrity-cases/<case-id>.md(cluster-level evidence sheets)
Analysis signals
- token-level similarity after normalization
- n-gram fingerprint overlap
- control-flow keyword profile similarity
- shared rare-error patterns when available
- metadata anomalies when available
Output semantics
integrity-report.json must include:
overall_status:pass|review|failpair_findings: suspicious pair list with scores and triggered signalsclusters: grouped suspicious submissionsmethod: scoring and thresholds usedpolicy_note: explicit statement that no automatic grade penalty is applied
Rules
- Never auto-penalize students based on similarity signals alone.
- Always provide evidence and alternative explanations.
- Exclude obvious boilerplate/template regions when possible.
- Require human attestation before disciplinary action.