scholar-evaluation

SUMMARY

What it does

This skill provides qualitative-first, evidence-traceable developmental review of scholarly works (papers, drafts, protocols, literature syntheses, research ideas) and audits low-stakes research-assessment rubrics. It includes optional local quality checks via bundled Python scripts (validate rubric, calculate scores, check traceability, summarize agreement, weight sensitivity, check process, generate report scaffold). It enforces a hard safety boundary: never use for hiring, promotion, tenure, admissions, grants, prizes, discipline, or any high-impact personnel decision; never rank people or reduce them to composite scores. It prohibits scoring based on metrics like impact factor, h-index, citations, or prestige. Data handling is strictly local: scripts accept only JSON/CSV with pseudonymous IDs and bounded ratings; no network, credentials, external models, or subprocesses. The skill references ScholarEval as an experimental framework, not validated psychometrics, and requires human review before releasing organizational reports.