What is shuxue-jianmo-lunwen-pingshen?

This project is a skill for AI coding assistants that reviews mathematical modeling competition papers against the standards used by China's national undergraduate mathematical modeling contest, the CUMCM. You give it a paper and it checks the paper's structure, format, and content the way a real competition judge would, then produces a scored report, a list of formatting problems, and specific suggestions for improvement. The skill first identifies which of four paper types the submission is, traditional modeling, data driven, machine learning, or engineering and signal processing, since each type gets evaluated somewhat differently. It checks whether the paper contains the eight standard sections a competition paper is expected to have, such as the abstract, problem restatement, model assumptions, and model solution, and flags any missing sections. It then reviews six formatting dimensions, including heading and paragraph structure, figure and table numbering, formula formatting, and reference citation style, recording the exact location and severity of every issue found. For content quality, it scores the paper across five weighted dimensions covering the reasonableness of the paper's assumptions, the creativity of the modeling approach, the clarity of the results, formatting compliance, and reference quality, giving each dimension a score along with a written justification rather than a bare number. Because judging genuine creativity is inherently subjective, the skill uses a list of innovation signals with a confidence rating, and automatically recommends human review for cases it cannot confidently judge itself. The project was tested against nineteen real competition papers and states it does not invent content or issue a final verdict on whether a paper will win an award, leaving that decision to human judges. It is installed by cloning the repository into a Claude Code or similar assistant's skills folder. The project uses a custom copyright license that allows personal use but forbids redistribution or commercial use without the author's written permission.