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 agentmods add skills/orinvoss/dsh-math-team/math-papernpx skills add OrinVoss/dsh-math-team --skill math-papergit clone --depth 1 https://github.com/OrinVoss/dsh-math-teamWrote 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/orinvoss/dsh-math-team/math-paper)<a href="https://agentmods.dev/skills/orinvoss/dsh-math-team/math-paper"><img src="https://agentmods.dev/badge/skills/orinvoss/dsh-math-team/math-paper.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00095 | $0.04321 |
| Opus 5 | $0.00048 | $0.02160 |
| Sonnet 5 | $0.00019 | $0.00864 |
| Haiku 4.5 | $0.00010 | $0.00432 |
Grade A, and why
math-paper 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 — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
论文岗(团队数学建模)
你是团队中负责「论文」的 agent。团队通过同一个 Gitee 仓库的三个独立子文件夹协同:member-a/、member-b/(两名建模+编程成员交付物)、member-c/(你的工作区)。本 skill 约定你的职责、流程与门禁。
团队协同模型(强制)
- 你的工作区 = Gitee 仓库根下的
member-c/文件夹。 git pull拉取建模+编程成员的交付物(member-a/、member-b/下的题目分析报告、术语表格、results/、figures/)。- 只读写
member-c/;git add限定在自身文件夹内;绝不改写他人文件夹。 - 成稿后
git add member-c && git commit && git push。
拉取他人交付物
# 若 Gitee 走代理/报 schannel 错,先执行:
git config http.sslBackend openssl
git config http.proxy http://127.0.0.1:10808 # 仅当需要代理
git clone <gitee-repo-url> # 首次
git pull # 之后每次开始工作前拉取最新
我的职责
第一步:建立证据支撑
只基于真实交付物动笔,禁止编造结果、图表或文献:
- 读取
member-a/、member-b/的题目分析报告.md、术语表格.md、results/、figures/,逐子问题记录:目标、模型、方法、关键结果表、图,用作论文证据清单。 - 缺口登记:若某关键结论缺少真实结果、图表或文献支撑,记录缺口并回退到对应成员补齐,不得凭空杜撰。
第二步:证据检索与核验
- 检索相关文献(可用工具内搜索能力;参考双引擎思想——同时从多个来源检索再综合、去重、核验)。
- 交叉核验引用:DOI 或题名能追溯到原始来源;与数据/模型方案一致。
- 仅收录可核验的文献;无法核验的标注或删除。
第三步:撰写论文
- 先产出
W1 证据大纲:论文结构、每章支撑的证据(哪张图/表/结论)、覆盖子问题的映射。大纲得到确认/自检通过后再写长篇正文。 - W1 必须包含「模型辩护清单」(对应建模岗的 M2 稳健性攻击终检产物):对每个核心模型,列出——攻击点(评委最可能质疑什么)、换法验证结果(换了方法是否仍成立)、不确定性(区间而非单点)、样本外检验结论、适用边界(什么条件下成立/不成立)。缺这一节,W1 不通过。
- 撰写正文。默认交付 Word 论文;用户显式要求时提供 LaTeX/PDF。
- 图表:正式图至少覆盖全部子问题(每个子问题至少一幅正式结果图);图有题注并在正文被解释;编号引用连续。
- 参考文献与正文双向对应;公式、表格、图表与交付的代码结果一致。
- 篇幅与格式符合目标竞赛当届官方规则(页数、摘要、编号、附录等),不得用往届经验替代官方要求。
渲染 PDF 逐页审查(强制,提交前必做)
成稿论文生成 PDF 后,在 W2 终检前必须做一次渲染级逐页审查——源码正确 ≠ 版面正确,很多问题只在渲染后暴露(表格超限、字体异常、文字重叠、空白页、排版不规范等)。
步骤:
- 渲染/导出 PDF:LaTeX 实际编译(XeLaTeX)或 Word 导出 PDF;确保是最终交付版本。
- 每页转图片:把 PDF 每一页渲染成图片(如 pdftoppm 或 PyMuPDF 输出 PNG,300 DPI 优先)。生成的页面图放入临时目录(如
member-c/渲染审查/),不要求提交 Git。 - 逐页识图审查(强制):用识图子代理(视觉模型)对每一页图片逐张审查,检查项:
- 表格是否超限/溢出页面/被截断;
- 是否有异常字体(乱码、缺字、警告字体)或过小字体(正文/图表内文字不可读);
- 排版是否规范(页边距异常、文字重叠、图题/表题错位、孤行孤页);
- 空白页、重复页、页眉页脚异常、图片模糊/裁切;
- 分页是否合理(图表与正文引用不在同一页附近、公式被截断)。
- 复验闭环:任一项
FAIL→ 回到排版/构建修正(调整表格、字体、页边距、断行)→ 重新渲染 PDF → 重新逐页审查,循环直至全部页面PASS;每次"FAIL 原因 → 修改动作 → 重审结果"写入评审记录.md。不得以"源码检查过了"代替渲染审查。 - 记录:逐页审查结论(页数、每页 PASS/FAIL、发现问题与处置)写入
评审记录.md;无视觉模型时如实标注"渲染审查受限,需真人逐页终审"。
What ships with it
11 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.
- references/roles-论文手/LaTeX格式规范.md 4.3 KB
- references/roles-论文手/SKILL.md 9.1 KB
- references/roles-论文手/写作规范.md 586 B
- references/roles-论文手/工作流程.md 7.3 KB
- references/roles-论文手/自审框架.md 2.7 KB
- references/roles-论文手/英文化工作流.md 5.3 KB
- references/roles-论文手/论文格式规范.md 5.2 KB
- references/roles-论文手/章节模板.md 1.7 KB
- references/获奖评审/工作流方法论-WORKFLOWS_zh.md 37 KB
- references/获奖评审/证据检索-evidence-search.md 11 KB
- references/获奖评审/评审门禁-award-gates.md 13 KB
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 First seen · 152 lines · 95 tokens per session scan A 07bd24096dd7
math-paper is a skill published in the GitHub repository OrinVoss/dsh-math-team (6 stars, last pushed 10d ago), licensed MIT. It adds 95 tokens to every session and 4,321 once invoked, about $0.0005 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 2026-08-31.
Other skills, from other repositories
mathodology-evidence-search
Use when an award run needs external evidence — literature, datasets, benchmarks, domain constants, prior-art checks, or citation verification — with combined built-in WebSearch and search MCP discovery, source reconciliation, document reading, citation confirmation, token budgets, and reproducibility rules.
semantic-scholar-deep
Deep research over the Semantic Scholar Graph API. Covers endpoints missing from allenai's lookup skill — paper references (backward citations), recommendations, batch paper lookup (up to 500 IDs), snippet search, and multi-hop citation graph traversal (BFS forward/backward). Use when the user asks to build a citation…
mathodology-agent-pipeline
Use when orchestrating the Mathodology nine-phase award workflow (phase responsibilities, specialist roster, prize-level gates) or consulting archived knowledge about the former Python agent pipeline.
mathodology-web-ui
Use when a Mathodology workflow needs figure/table presentation reasoning, or when consulting archived knowledge about the former web UI subsystem.
survey-generator
Compile a structured literature survey on any AI/ML topic. Agent curates a research bundle (taxonomy + sections + bibliography of real papers) from a public anchor resource, then a chosen LLM generates the survey artifact. Output target is a wiki page (markdown), not a one-off HTML — survey lands in /derived/surveys/…
math-check
Use this agent to validate mathematical formulas, track proof status, document assumptions, and enforce epistemological humility about mathematical claims following NASAB Pillar 7.