What problem does it solve? Mathematical modeling competition papers increasingly show signs of AI-generated content, but reviewers lack a transparent, quantified method to detect it. This Skill applies a two-layer audit framework to score AI-style risk in Chinese modeling papers and outputs actionable HTML reports. ## Core Features & Use Cases - Two-layer scoring framework: Layer 1 scans language templating (high-frequency connectives, parallel sentences, passive voice, unsupported superlatives) weighted at 60% plus four other dimensions; Layer 2 reviews modeling logic including model commonality, complexity-vs-benefit, five vanity patterns, and a three-question justification method. - Quantified risk output: Computes a 0-100 AI risk score with five risk tiers, parameter traceability tables, and P0/P1/P2 prioritized revision suggestions. - Layered HTML reports: Generates a concise self-check version for students and a complete review version for instructors and judges. - Use Case: A competition advisor receives a student paper, runs the audit, and gets a scored report pinpointing unsupported claims, unjustified model choices, and untraceable parameters with concrete rewrite suggestions. ## Quick Start Audit this mathematical modeling paper for AI-generated traces using the two-layer framework and generate the HTML risk report.
