Lucasuiii/modeling-workbench / .agents/skills/cumcm-workflow
cumcm-workflow
VerifiedBuild or resume mathematical-modeling competition work, including CUMCM, MCM/ICM, graduate and regional contests, data challenges, and open-topic statistical modeling. Guide problem framing, modeling, computation, validation, Chinese or English LaTeX writing, and reviewed PDF/source delivery using current official requirements. Not for ordinary paper polishing.
references/competition-adaptation.md
View on GitHub# Adapt to the competition, preserve the workflow
For Huawei Cup / 中国研究生数学建模竞赛, also read [Huawei delivery adaptation](huawei-delivery.md): current official cover, abstract/anonymity checks, submission filenames, attachments and measured MD5 binding.
Use this guide when starting a competition other than a familiar CUMCM problem, or when the requested language, deliverables or research format differ. It is a routing guide, not a source of current competition rules. The same competition can contain several task families.
## Identify the actual assignment
Read the user-supplied current official problem/theme, attachments and submission instructions using [intake](01-intake.md). Distinguish a paper template, a results spreadsheet and a document describing submission rules; their file extensions do not determine their roles.
Summarize the competition/year, requested language, fixed problem versus open topic, required outputs and material uncertainties in the existing `PROJECT_BRIEF.md`. Keep substantive task requirements in `PROBLEM_FACTS` and `TASK_CAPABILITIES`, with source locations, so they reach modeling and review. These are facts and writing requirements, not new runtime options. A note in the brief does not reconfigure a script.
For submission constraints, retain the source location and precise scope: which pages count, which attachments are separate, where identifiers are required or forbidden, accepted formats, and any restrictions on data or tools. Unknown is not permission and not a guessed default. Ask only about material missing information; continue reversible analysis where it is unaffected. Never fetch neighboring official files without user authorization or replace current rules with a competition-name lookup.
## Route by the work, not by a template count
The examples below identify useful entry points, not verified support for each organizer/year.
| Assignment encountered | Examples of competition labels | Guidance to load |
|---|---|---|
| Fixed problem, Chinese modeling paper | CUMCM, 华为杯, MathorCup, 电工杯, 华东杯, 华中杯, 五一杯, 深圳杯 and other regional contests | Existing stages; select sections of [task-driven modeling](task-driven-modeling.md) from the actual mathematical task |
| Fixed problem, English paper and possibly an audience-specific document | MCM/ICM, APMCM, international 数维杯 or 认证杯 tracks | Same modeling/validation stages; [competition writing](competition-writing.md) for English and supplementary documents |
| Data-centered prediction, classification or decision task | 泰迪杯 and data-oriented tasks in other contests | Data audit and predictive evaluation in [task-driven modeling](task-driven-modeling.md); preserve required result-file layouts |
| Official theme with a team-selected research question | Statistical-modeling and open-topic tracks | [Open-topic research](open-topic-research.md), then the common model/computation/validation stages |
| Engineering, physical or simulation-heavy task | Any of the above | Mechanism, units, numerical accuracy and parameter identification; do not pick a solver merely from the competition label |
No fixed model count, chart quota, page target or diagram tool follows from this table. A topic named “prediction” may actually require an actionable decision; an English-language problem may require no memo at all.
## Shared LaTeX/PDF delivery
- `init_project.py --official` accepts a supplied official theme/rule document as an input; it does not require a pre-existing numbered problem. Supply a truthful project ID, keep team data separate from the official input set and classify provenance honestly.
- After the existing conclusion checkpoint, `init_latex_paper.py --competition <actual-name> --language zh|en` generates a Chinese CTeX or English article scaffold. The competition name is stored in the template manifest. Default `--template auto` selects the editable Huawei typography preset for recognized Huawei/graduate modeling names in Chinese; other contests keep the generic scaffold. `--template generic` overrides that selection. Typography selection does not certify current official rules. Omitting the options preserves CUMCM/Chinese defaults. The template manifest keeps the chosen scaffold ID/mode, so downstream tools use the actual files rather than guessing language from the contest name.
- Use the same `record_compile.py`, `paper-delivery` handoff, package builder and final checks for every competition. They bind the real PDF and editable sources. The competition is part of the existing paper-stage snapshot and handoff; changing it after review requires renewed review through the existing process.
- A declared official paper template still takes priority: generic initialization stops so the agent can adopt/adapt the supplied template and maintain the existing `official_package_adapter` manifest. That adapter may now identify the actual competition. Automated conversion of arbitrary official templates is not provided.
- Generic scaffolds start with `official_compliance: unverified`. Apply current rules for title/summary sheets, page size/counting, identifiers, appendices and required supporting files before the existing final compliance review. A successful compile does not establish compliance.
- Delivery expects a reviewed PDF, editable LaTeX sources and computation sources. Word/Markdown-only delivery has no export-receipt path. Do not rename a DOCX ZIP as a LaTeX package or mark an ordinary converter run as a TeX compile.
For example, after `PAPER_PLAN` and current human conclusion acceptance are ready:
```bash
python3 "$S/init_latex_paper.py" --project <p> --competition 'MCM/ICM' \
--language en --competition-year <year> --title '<actual title>' --keywords '<actual keywords>'
```
Use `--language zh` with the actual competition name for a Chinese paper. Read [LaTeX template guidance](latex-template.md) for the shared compile/render process. Synthetic pipeline tests and real scaffold compilation exercise the infrastructure; they do not certify every organizer/year or replace real-problem validation.
## Preserve the current decisions
Present the selected research scope, objective, candidates and evidence at the existing model-design checkpoint. Present reviewed conclusions before writing. Retain the final-artifact checkpoint for any actual delivery. There is no extra mandatory “competition approval” or “topic approval” stage.
If a rule clarification changes the task, model or accepted conclusions, use the existing revision and re-review procedure for the affected work. Do not treat a prose edit in the brief as automatically invalidating machine snapshots; put consequential changes into the relevant existing contracts and explicitly reopen affected decisions. Do not claim new automatic rule checking: semantic completeness and current-rule compliance still require review.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| canvas-designanthropics/skillsCreate beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| docxanthropics/skillsUse this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx files) or Word templates (.dotx files). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, headings, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, performing find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file, use this skill. Do NOT use for PDFs, spreadsheets, Google Docs, or general coding tasks unrelated to document generation. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| pdfanthropics/skillsUse this skill whenever the user wants to do anything with PDF files. This includes reading or extracting text/tables from PDFs, combining or merging multiple PDFs into one, splitting PDFs apart, rotating pages, adding watermarks, creating new PDFs, filling PDF forms, encrypting/decrypting PDFs, extracting images, and OCR on scanned PDFs to make them searchable. If the user mentions a .pdf file or asks to produce one, use this skill. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| pptxanthropics/skillsUse this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an email or summary); editing, modifying, or updating existing presentations; combining or splitting slide files; working with templates (.potx), layouts, speaker notes, or comments. Trigger whenever the user mentions "deck," "slides," "presentation," or references a .pptx or .potx filename, regardless of what they plan to do with the content afterward. If a .pptx or .potx file needs to be opened, created, or touched, use this skill. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |