What problem does it solve? Research teams accumulate large paper portfolios without knowing which outputs are near publishable quality and which should be revised or archived. This Skill produces a cross-validated evaluation report where every number is traceable to source data, preventing decisions based on single scores or fabricated statistics. ## Core Features & Use Cases - Multi-Dimensional Cross-Validation: Combines quality gate status (PASS/CONDITIONAL/FAIL), tier grades (T1–T4), and D10a citation coverage instead of relying on a single quality score. - Compound-Threshold Filtering: Identifies the small set of core assets meeting strict criteria (T1+T2 and D10a≥95%) and separates factual measurements from inferred estimates. - Disposition Classification & Data Honesty: Labels each low-quality output as needing major revision or archival, counts zombie citations per paper without folding them into means, and rejects unsourced numbers with recovery suggestions. - Use Case: Given a directory of 98 papers with mixed .tex/PDF availability and quality scores, produce a report showing gate distribution, missing-data statistics, the 10–15 near-publishable papers, and per-paper disposition labels. ## Quick Start Evaluate this directory of research papers and produce a multi-dimensional quality report with gate distribution, core asset filtering, and per-paper disposition labels.
