agentleFS
LocloveLoclove/cumcm-skills/skills/cumcm-paper-writing/SKILL.md

cumcm-paper-writing

Skill4 starsChanged 3 months agoMITView on GitHub

Draft, restructure, and audit CUMCM mathematical-modeling papers with a default 25--30-page body, Song-style Chinese typography, Nature-style claim-evidence-boundary writing, compliant sections, model equations, citations, data references, integrated figures/tables, appendix lists, and final formula rendering. Use when the user asks to write the abstract, model sections, assumptions, symbols, results, sensitivity analysis, conclusion, references, appendix, or a full contest paper.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 10 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write one bounded argument sentence: problem and scope -> advance -> approach -> evidence -> boundary.

    line 14
  2. Distinctive:

    Write the abstract last, after the final model route and results are fixed.

    line 19
  3. Distinctive:

    Ensure at least 8 literature/data citations when external sources materially support the paper.

    line 20
  4. Distinctive:

    Write formulas in standard LaTeX display math and cite copied/adapted formulas at first use.

    line 21
  5. Distinctive:

    Keep all Chinese text, captions, and figure labels in Song-style fonts under the enhanced profile.

    line 22
  6. Very specific:

    Run scripts/check_citations_formulas.py before final handoff, then hand off to cumcm-reviewer and cumcm-format-packager.

    line 23
  7. Distinctive:

    Do not include a目录 when following the 2026 format constraint.

    line 43
  8. Distinctive:

    Do not deliver an enhanced-profile body below 25 pages or above 30 pages.

    line 44
  9. Distinctive:

    Do not reach 25 pages through repeated background, oversized layout, redundant figures, split equations, or unsupported expansion.

    line 45
  10. Very specific:

    Do not use SimHei/YaHei/Kaiti/FangSong as casual Chinese fallbacks under the Song-font profile.

    line 46
  11. Distinctive:

    Do not let formulas become screenshots unless the final Word/PDF workflow requires it and a formula ledger remains auditable.

    line 47
  12. Distinctive:

    Do not write unsupported superlatives such as 最优, 显著, 精确, 鲁棒 without evidence.

    line 48

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

CUMCM Paper Writing

This skill turns model work into a concise, traceable, contest-ready paper.

Workflow

  1. Load ../cumcm-shared/references/format-spec-2026.md, ../cumcm-shared/references/enhanced-writing-profile.md, and ../cumcm-shared/references/citation-formula-standards.md.
  2. Build and lock a terminology ledger for model names, data, variables, metrics, units, abbreviations, and notation.
  3. Write one bounded argument sentence: problem and scope -> advance -> approach -> evidence -> boundary.
  4. Build a paper skeleton and 25--30-page budget from actual subquestions and scored deliverables. Do not force a generic template when the problem structure differs, and do not pad to reach the minimum.
  5. Give each paragraph one job and draft from evidence outward. For each subproblem use: task/scope -> baseline -> main model and rationale -> solution -> result -> fair comparison -> robustness/sensitivity -> failure boundary -> transition.
  6. Build a claim-evidence table: every sentence claiming improvement, optimality, prediction, robustness, or policy value must point to a model output or citation.
  7. Build a figure integration ledger before drafting around figures. Introduce the question/claim before each figure, place the figure near first mention, and interpret its trend, mechanism, and boundary immediately after it.
  8. Write the abstract last, after the final model route and results are fixed.
  9. Ensure at least 8 literature/data citations when external sources materially support the paper.
  10. Write formulas in standard LaTeX display math and cite copied/adapted formulas at first use.
  11. Keep all Chinese text, captions, and figure labels in Song-style fonts under the enhanced profile; use bold Song rather than switching heading families unless a current regional template requires otherwise.
  12. Run scripts/check_citations_formulas.py before final handoff, then hand off to cumcm-reviewer and cumcm-format-packager.

Default Paper Skeleton

摘要
关键词
一、问题重述
二、模型假设
三、符号说明
四、数据处理与探索性分析
五、模型建立与求解
六、结果分析、检验与敏感性分析
七、模型评价、推广与结论
参考文献
附录

Red Lines

  • Do not include a目录 when following the 2026 format constraint.
  • Do not deliver an enhanced-profile body below 25 pages or above 30 pages; count from electronic page 2 to the page before the appendix and manually review a mixed body/appendix page.
  • Do not reach 25 pages through repeated background, oversized layout, redundant figures, split equations, or unsupported expansion.
  • Do not use SimHei/YaHei/Kaiti/FangSong as casual Chinese fallbacks under the Song-font profile.
  • Do not let formulas become screenshots unless the final Word/PDF workflow requires it and a formula ledger remains auditable.
  • Do not write unsupported superlatives such as 最优, 显著, 精确, 鲁棒 without evidence.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.