Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add haoxilin/cumcm-live-workflow-skill --skill cumcm-live-workflowgit clone --depth 1 https://github.com/haoxilin/cumcm-live-workflow-skillWrote 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/haoxilin/cumcm-live-workflow-skill/cumcm-live-workflow)<a href="https://agentmods.dev/skills/haoxilin/cumcm-live-workflow-skill/cumcm-live-workflow"><img src="https://agentmods.dev/badge/skills/haoxilin/cumcm-live-workflow-skill/cumcm-live-workflow/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/haoxilin/cumcm-live-workflow-skill/cumcm-live-workflow"><img src="https://agentmods.dev/badge/skills/haoxilin/cumcm-live-workflow-skill/cumcm-live-workflow.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 | $0.00223 | $0.09801 |
| Opus 5 | $0.00112 | $0.04901 |
| Sonnet 5 | $0.00045 | $0.01960 |
| Haiku 4.5 | $0.00022 | $0.00980 |
Grade A, and why
cumcm-live-workflow 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 4d 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 — 335 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CUMCM 真题全流程实战手册
沉淀自多轮真实竞赛实战、交叉审阅与 AI 自查表终审整改，聚焦可执行的流程、坑位与质量门禁，通用适配任意 CUMCM 题目。
触发条件
启动关键词（任一命中即启动本技能）：
- 流程类：完整做真题 / 按竞赛标准走完流程 / 数学建模竞赛实战 / 打国赛 / 备赛训练 / 模拟竞赛
- 环境类：环境检查 / 检查环境 / 准备环境 / 环境就绪
- 终审类：按 AI 自查表审查 / 自查表终审 / 论文终审整改
具体场景：
- 用户开始完整做一道真题（练手或正式比赛）
- 备赛训练：模拟完整流程
- 已出题、要求"按竞赛标准走完流程"
- 拿到成品论文，要求按 AI 自查表终审整改
- 启动后第一步：自动执行 §0 环境检查并报告完成度，再进入读题
工作区约定
- 题目材料放
PROJECT_ROOT/<题号>/：原题 PDF、格式规范 doc、附件/ - 建模产物（代码/npz/图）放题号目录；论文放
题号/paper/；结果 xlsx 放题号目录 - 附件模板只读：填写时 load 原模板写单元格另存，不覆盖原附件
流程总览（按序执行）
0. 环境检查 → 1. 读题+格式规范 → 2. 逐问建模+出图（至少8图）
→ 3. LaTeX 论文 → 4. 交叉审阅（≥2轮）→ 5. 填 result 模板
→ 6. AI 自查表驱动终审 → 7. 清理+打包
0. 环境检查（启用后自动执行，无需用户要求）
0.1 自动检测并报告完成度（先做，不询问）
技能一经启用，立即后台探测本地环境，输出前置条件完成度清单：
| 组件 | 检测命令 | 通过标准 |
|---|---|---|
| Python | python --version |
≥ 3.12 |
| numpy / scipy / pandas / matplotlib / openpyxl | python -c "import numpy,scipy,pandas,matplotlib,openpyxl" |
全部可导入 |
| scikit-learn、black（可选） | 同上 | 可导入 |
| LaTeX（xelatex） | xelatex --version |
MiKTeX 或 TeX Live 可用 |
| ctex 宏包 / 中文字体 | kpsewhich ctex.sty + SimSun/SimHei 存在 |
通过 |
| pdftotext / pdftoppm（可选） | --version |
通过 |
报告格式示例：
[环境检查] Python 3.12.10 ✓ | numpy ✓ | scipy ✓ | pandas ✓ | matplotlib ✓ | openpyxl ✓
LaTeX (xelatex) ✗ 未安装 | ctex 宏包 ✗ | pdftotext ✗（可选）
完成度：6/9 ✓ 必装组件缺失：LaTeX、ctex
0.2 确认后自动安装缺失组件
把清单交给用户确认（"缺失组件是否自动安装？"），得到明确确认后执行：
- pip 组件：
pip install X -i https://pypi.tuna.tsinghua.edu.cn/simple（国内镜像） - LaTeX（xelatex + ctex）：优先
winget install MiKTeX.MiKTeX（或提示用户用官方安装器 / TeX Live）；装后需刷新 PATH 或重开终端 - poppler（pdftotext / pdftoppm）：
winget install poppler（可选组件，缺失可跳过，不阻塞） - 安装完成后复测，更新完成度清单直至 100% 或用户手动处理；最终报告"环境就绪，可开始读题"
- 红线：未获用户确认不得安装任何软件；涉及系统级变更（装 LaTeX/驱动等）时给出明确提示
- 网络：pip 走清华镜像；若网络受限（如需代理），先探测
HTTPS_PROXY环境变量并询问用户
1. 读题 + 格式规范
题意理解优先（最高优先级教训，任何题目通用）
- 题干明写的物理事实必须建模进去：题面给出物体有宽度/厚度/形状 → 碰撞、干涉、间隙类判定必须用其真实几何（矩形/多边形/角点-中心线距离判据），严禁简化为"中心线/质点间距"判据——后者会漏检真实干涉，导致终止时刻、临界参数等关键结果整体失真
- 几何初始条件必须用解析公式独立验证，禁止只信单一来源：凡能用公式从题干数字推出的参数（如半径=螺距×圈数、长度=孔距+2×孔心距、角度=弧长/半径），一律推导核对一遍，不能只信视觉读图
- 后问对前问误差高度敏感：初始条件错→中间判据错→后续结果连锁全错；问题间耦合必须每步验证
- 无官方答案时的题目自查（真实比赛核心），读完题后逐条执行：
- 题干数值→解析公式独立推导：把题面每个数字代入公式算一遍，凡是能算出来的参数不许依赖读图/猜
- 单位与几何自洽检查：所有长度/时间/速度换算成同一单位系；画示意图核对几何关系
- 边界情形推演：取极端参数（最大/最小）走一遍模型，看结论是否符合物理直觉——能抓出大部分"模型没建对"
- 多方法交叉验证：同一结论至少两种独立途径（解析式 vs 数值扫描 vs 特例验证），一致才算数
- 把"题面明写的物理事实"列清单：写进代码注释/文档，建模时逐条对照是否全部体现
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago Changed · +7 lines 2fe23ff2384d
- 12d ago First seen · 328 lines · 223 tokens per session scan A a0617996ea53
cumcm-live-workflow is a skill published in the GitHub repository haoxilin/cumcm-live-workflow-skill (146 stars, last pushed 4d ago), licensed MIT. It adds 223 tokens to every session and 9,801 once invoked, about $0.0011 per session on Opus 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.
Other skills, from other repositories
1start-mathmodel
A workflow entry point for the Chinese national mathematical modelling competition. It coordinates problem analysis, modelling, programming, diagrams, paper writing, and verification.
review
Write domain-authentic review articles that synthesize rather than stack. Covers 10 scientific domains (physics, chemistry, biology, medicine, mathematics, computerscience, earthenvironment, astronomy, economics, materials) with per-domain style guides mined from landmark reviews in RMP / Nat. Rev. X / Annu. Rev. /…
matplotlib-figures
Publication-quality data plots via matplotlib with venue-specific styles. Use for generating your own figures (timelines, comparison charts, data summaries, heatmaps) that you add to the LaTeX report. Your brain prompt supplies the venue-specific style directory as {{VENUESPECIFICDIR}} — use that value wherever this…
qec-construct
Verifier-in-the-loop CONSTRUCTION of quantum error-correcting codes with transversal non-Clifford gates (CCZ/T). Applies whenever the project goal is a new or better code/construction — INCLUDING search-phrased goals ("find codes beating X"), where the construct-loop (propose algebraic rule → qverify → debug) is the…
figure
Hybrid figure pipeline (Nano Banana raster + rembg background removal + TikZ vector assembly). Includes a TikZ template library covering quantum circuits (quantikz), Feynman diagrams (tikz-feynman), circuits (circuitikz), molecules (chemfig), 2D/3D plots (pgfplots), energy-level diagrams, phase-space trajectories…
paper-figures
Extract figures from downloaded papers and include them in survey/review reports. Use when the report covers other groups' work and benefits from their architecture diagrams, experimental plots, or system schematics. Your brain prompt supplies the absolute path to the extract-figures script as {{EXTRACTFIGURES}} — use…