Math Modeling Skills
Two skills for the CUMCM and MCM/ICM math modeling contests: problem breakdown, model choice, code and the paper
- 428 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Python and MATLAB templates run on your machine, like any agent-generated code
- At the literature stage the skill runs up to five web searches, and parts of the problem statement may end up in queries
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/math-modeling-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Python and MATLAB templates run on your machine, like any agent-generated code
- At the literature stage the skill runs up to five web searches, and parts of the problem statement may end up in queries
This is third-party code. Review the repository files before installing.
What it does
Math Modeling Skills is a pair of skills for math modeling contest problems: China's CUMCM (problems A, B, C) and the international MCM/ICM (problems A–F). The author derived the rules and templates from an analysis of more than 150 award-winning papers from 2018–2025. The math-modeling-solver skill leads from the problem statement to code. It classifies the mathematical nature of each subproblem (12 types in all), searches literature when needed, picks a model from a matrix of 95 scenarios with alternatives and applicability limits, and supplies code. The set has eight algorithm cookbooks, 22 Python templates, 7 MATLAB templates and 12 worked examples covering the full path from problem to solution. The math-modeling-paper skill helps write the paper: a structure for both contests, abstracts in Chinese and English with samples, sensitivity and robustness checks, figure preparation, a library of academic phrasing and Memo and Letter formats for MCM/ICM. Separate sections clean out signs of AI writing and run a four-round self-review. The solver hands material to the writer as a fragment tagged [PAPER_READY].
Who it is for. For students and teams preparing for CUMCM and MCM/ICM, and for teachers who walk students through such problems.
Good fit when
- You received a CUMCM or MCM/ICM problem and do not know how to split it into subproblems and identify the model type
- You need to pick a model for a scenario and compare candidates, such as XGBoost and random forest, with the limits of each
- You need working code for a genetic algorithm, particle swarm, simulated annealing or Monte Carlo from ready Python and MATLAB templates
- The model is done and you must write the abstract, build the paper structure and a literature review from award-winning samples
- An English draft should go through a four-round review with signs of AI writing cleaned out
Not a fit when
- The task is not math modeling: the skills are built for breaking down contest problems
- Your contest or university rules restrict AI assistance: responsibility for compliance stays with you
- You want a ready answer with no choices of your own: the matrix recommendations are a starting point, and the team chooses and justifies the model
- You are writing in a language other than Chinese or English: samples and phrasing are prepared only for those
Example request
Here is the statement of MCM problem B. Break it into subproblems, identify the model type for each and suggest two candidates with their applicability limitsLimitations
- Instructions and reference files are written in Chinese. Samples and templates are prepared for CUMCM and MCM/ICM.
- Installing is more than cloning the repository: Claude Code scans only folders shaped like ~/.claude/skills/<name>/SKILL.md, so the skills/ subfolders are installed through `npx skills add` or moved by hand. Restart Claude Code afterwards.
- The literature search stage is capped: no more than five web searches and 5–8 found papers, after which the search stops. It can be skipped.
- Model recommendations draw on an analysis of award-winning papers and serve as a starting point for your own choice.
- For figures and documents the author suggests pairing separate skills: nature-figure, nature-polishing, xlsx, docx and pdf. They are not part of the set.
How to disable. Delete the math-modeling-solver and math-modeling-paper folders from ~/.claude/skills/, or remove them through the same `npx skills` tool you installed with. Restart Claude Code.
FAQ
How do I install Math Modeling Skills in Claude Code?
Run `npx skills add Lupynow/math-modeling-skills --skill '*'` to install both skills. You can install just one, for example `npx skills add Lupynow/math-modeling-skills --skill math-modeling-solver`. Restart Claude Code afterwards.
How is math-modeling-solver different from math-modeling-paper?
The solver leads from the statement to a model and code: it breaks the problem down, searches literature when needed, picks a model and expands the algorithm. The paper skill takes the finished model and helps with the text, figures and checks. The [PAPER_READY] tag sits between them.
Which contests does the set cover?
China's CUMCM with problems A, B and C and the international MCM/ICM with problems A–F. MCM/ICM also has dedicated material: model naming, Memo and Letter formats and the Our Work diagram.
Which code templates are included?
22 Python templates and 7 MATLAB templates. The cookbooks cover optimization, machine learning, evaluation, mechanistic models, statistics, networks, clustering and game theory.
How many web searches does the literature stage run?
No more than five: two or three in English and two or three in Chinese, stopping after 5–8 papers are found. The stage is recommended but skippable, for example when you have already gathered sources.
Discussion
No questions or reviews yet.
Related
SkillsA three-stage workflow for math modeling tasks: analysis and model choice, Python or MATLAB computation, and paper assembly in Word or LaTeX
A skill set for math modeling contests: problem parsing, method selection, Python and MATLAB code, and paper writing
Eight skills for math modeling contests that carry a task from the problem to a paper with figures and checked references
An agent skill that turns an arxiv paper into a minimal Python implementation, with code anchored to sections and equations and gaps flagged honestly
DDC Skills for Construction
DDC Skills for AI Agents in Construction
238 skills for construction companies: BIM to Excel export, cost item search, estimates, site reports and document generation
A library of agent skills for medical research: literature search and appraisal, study design, data analysis and academic writing
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| Math Modeling Skills (this entry) | no | 428 | |
| Academic Writing Skills | no | 525 | eight skills for papers and theses in LaTeX and Typst, while Math Modeling Skills targets modeling contest papers and adds problem breakdown and code |
| Research Writing Skill | no | 531 | 30 prompt templates for scientific papers, while Math Modeling Skills offers solution and code templates plus a separate skill for the contest paper text |
README in short
The page presents the two skills with capability tables, then explains two install methods and the workflow from problem to submission. Below are a file tree describing the cookbooks, a list of suitable external skills, the license and a version log with details for each release.
README badge
Original SKILL.md
--- name: math-modeling-solver description: 数学建模竞赛解题全流程指导。覆盖国赛(CUMCM)和美赛(MCM/ICM)全部题型(A-F)，提供12种问题本质分析、95+场景模型决策矩阵、5本算法Cookbook、11本完整例题Playbook、22个Python+7个MATLAB可运行代码模板。与math-modeling-paper形成"解题→写作"配对。当用户提及建模思路、选什么模型、怎么建模、赛题求解、粘贴赛题文本、美赛/国赛题目分析、CVaR/NSGA-II/Monte Carlo/时间序列/ANOVA/灰色关联、网络流/图论/生态建模、模型命名/Memo/Letter/Our Work流程图时，使用此skill。 --- # 数学建模竞赛解题指导 > 本 skill 提供的矩阵、cookbook、playbook、代码模板，全部是**知识参考**而非**决策指令**。 > 对于同一道赛题，不同队伍理应有不同的建模路径。矩阵里的推荐只是技术起点—— > 你的任务是结合题目具体约束、数据特征和团队判断，做出**有理由的选择**，而非照搬推荐。 ## 使用流程 收到解题任务后，按以下五阶段工作流操作。 ### Step 0: 判断用户入口 先判断用户在哪个阶段切入： - **有新题目文本，从零开始** → 阶段1 - **已有问题分析结果，需要模型推荐** → 阶段2 - **需要先查文献再看选什么模型** → 阶段1.5 - **已确定模型，需要算法展开和代码** → 阶段3 - **建模已完成，需要衔接论文** → 阶段4 - **用户直接指定了模型名**（如"用 GA 求解"）→ 阶段3，跳过阶段1-2 - **用户输入匹配已知题型** → 加载对应 Playbook 获取完整解题示范（12 本 Playbook 覆盖国赛 A/B/C + 美赛 D/E/F 全部题型） ### 阶段识别规则 | 用户说 | 切入阶段 | |--------|---------| | "这道题怎么做" + 粘贴题目 | 阶段1 | | "帮我分析这道题" | 阶段1 | | "帮我搜一下类似问题的文献" | 阶段1.5 | | "有没有类似题目的论文可以参考" | 阶段1.5 | | "选什么模型好" | 阶段2 | | "XGBoost 和随机森林怎么选" | 阶段2 | | "帮我写 GA 代码" | 阶段3 | | "这个公式怎么推导" | 阶段3 | | "建模做完了，准备写论文" | 阶段4 | | "帮我写摘要" | 阶段4 → 引导切换 paper skill | --- ### 阶段1：拆题分析