| name | cumcm-support-material-audit |
| description | Audit CUMCM supporting materials for required code, data, generated outputs, AI-use details, appendix file lists, package size, cache files, identity leaks, and paper-to-file traceability. Use when preparing ZIP/RAR support packages or checking whether materials can reproduce paper results. |
CUMCM Support Material Audit
Use this skill before packaging supporting materials.
Workflow
- Identify required materials from the paper: code, data, generated figures, result tables, AI-use details, and large intermediate outputs.
- Compare the appendix file list with actual files.
- Check package risks: cache files, local paths, credentials, identity names, missing README/run notes, and oversized archives.
- Ensure paper claims can be traced to included outputs or reproducible scripts.
- Run
python3 scripts/check_submission.py --target cumcm --support-dir <dir> when a support directory exists.
- Read
references/checklist.md for package contents.
- Read
references/package-risks.md when classifying findings.
- Read
references/examples.md when writing final package notes.
Guardrails
- Do not include private or school-identifying files.
- Do not delete raw data without approval.
- Do not claim reproducibility when scripts require missing local paths or unavailable files.