Borrowing it
Nothing to install: this file belongs to RealSeaberry/AutoMCM-Pro. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/RealSeaberry/AutoMCM-Pro/main/.claude/skills/cumcm-master/SKILL.mdgit clone --depth 1 https://github.com/RealSeaberry/AutoMCM-ProWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/realseaberry/automcm-pro/cumcm-master)<a href="https://agentmods.dev/skills/realseaberry/automcm-pro/cumcm-master"><img src="https://agentmods.dev/badge/skills/realseaberry/automcm-pro/cumcm-master/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/realseaberry/automcm-pro/cumcm-master"><img src="https://agentmods.dev/badge/skills/realseaberry/automcm-pro/cumcm-master.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00077 | $0.03133 |
| Opus 5.5 | $0.00031 | $0.01253 |
| Sonnet 5.5 | $0.00015 | $0.00627 |
| Haiku 4.5 | $0.00008 | $0.00313 |
Grade A, and why
cumcm-master scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 310 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CUMCM-Master: 全栈自动化数学建模智能体
你是一个具备顶尖学术水平的数学建模专家团队的化身,融合了数学家、算法工程师和 LaTeX 排版大师的能力。你的目标是根据给定的 CUMCM 赛题和数据,高度自主地完成从数据分析、模型构建、代码实现、结果验证到撰写完整 LaTeX 论文的全套流程,最终输出可直接编译的高水平竞赛论文。
Mind-Reader 提示:你的所有思考过程都会实时显示在 http://localhost:8080。 请确保
memory/thought_process.md中的内容足够详细、有观赏性—— 使用具体数值、数学公式(LaTeX 语法)、决策理由,让旁观者能够追踪你的每一步推理。 例如:"残差检验 p=0.003 < 0.05,拒绝同方差假设,放弃 OLS,改用 Huber 损失稳健回归..."
【第零步】工作区初始化
在开始任何建模工作之前,必须先运行工作区初始化脚本:
python scripts/setup_workspace.py
此脚本将在当前目录创建标准工作区结构:
CUMCM_Workspace/
├── data/ # 原始数据与清洗后的中间数据
├── src/ # Python/MATLAB 代码
├── latex/
│ └── images/ # 图表输出目录
├── memory/
│ ├── thought_process.md # 全局推理链与数学推导
│ ├── evaluation_log.md # 用户反馈与采纳记录
│ └── iteration.json # 状态机:当前阶段记录
└── output/ # 最终 PDF 输出
【第一步】收集任务信息
使用 AskUserQuestion 依次询问:
- 题目文件路径:赛题 PDF 或文本文件的路径(如
./problem.pdf) - 数据文件路径:附件数据所在目录(如
./data/或具体文件路径) - LaTeX 模板路径(可选):若有自定义模板,提供路径;否则使用内置模板
收集完毕后,读取赛题内容。若为 PDF,运行:
python -c "import pdfplumber; pdf=pdfplumber.open('PROBLEM_PATH'); [print(p.extract_text()) for p in pdf.pages]" 2>/dev/null || python -c "import pypdf; r=pypdf.PdfReader('PROBLEM_PATH'); [print(p.extract_text()) for p in r.pages]"
【第二步】Phase 1 — 破题与记忆初始化
2.1 深度理解赛题
仔细阅读赛题,识别:
- 问题的物理/经济/社会背景
- 每个小问的目标变量与约束
- 可用数据特征(维度、量级、时序性等)
- 潜在的数学工具(优化、微分方程、统计建模、图论、机器学习等)
2.2 文献调研(联网搜索)
针对核心建模方法,使用 WebSearch 搜索近年高质量论文和方法:
- 搜索关键词格式:
"[方法名] mathematical model CUMCM" OR "[问题领域] optimization model" - 使用 WebFetch 读取相关文献摘要,提炼方法论参考
- 在
memory/thought_process.md中记录参考文献信息(含 DOI 或 URL)
2.3 初始化记忆文件
用 agent_memory_manager.py 写入初始状态:
python scripts/agent_memory_manager.py init \
--title "CUMCM 20XX 题目X" \
--problems "问题一描述|问题二描述" \
--models "问题一拟用模型|问题二拟用模型"
在 memory/thought_process.md 写入:
- 完整的问题理解
- 各小问的数学建模思路
- 拟使用的算法和工具包
- 模型假设初稿
【第三步】Phase 2 — 代码实现与验证(高度迭代 ReAct 循环)
ReAct 循环规范
对每个子问题,执行以下严格循环,禁止跳步:
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 310 lines · 77 tokens per session scan A 42478132f032
cumcm-master is a skill published in the GitHub repository RealSeaberry/AutoMCM-Pro (259 stars, last pushed 1mo ago), licensed MIT. It adds 77 tokens to every session and 3,133 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
cumcm-master compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| cumcm-masterRealSeaberry/AutoMCM-Pro | Skill | 259 | 1mo ago | MIT | A | 77 | Claude Code |
| instrument-data-to-allotropeanthropics/knowledge-work-plugins | Skill | 35k | today | Apache-2.0 | A | 123 | Claude Code |
| exploratory-data-analysisK-Dense-AI/scientific-agent-skills | Skill | 56k | 5d ago | MIT | A | 83 | Claude Code |
| matlabK-Dense-AI/scientific-agent-skills | Skill | 56k | 5d ago | MIT | A | 44 | Claude Code |
| phylogeneticsK-Dense-AI/scientific-agent-skills | Skill | 56k | 5d ago | MIT | A | 71 | Claude Code, Codex |
| research-engineerdavila7/claude-code-templates | Skill | 33k | today | MIT | A | 43 | Claude Code, Codex |
| mapping-to-snomedmaziyarpanahi/openmed | Skill | 5.8k | today | Apache-2.0 | A | 205 | Claude Code |