Mathodology skill
Eight skills for math modeling contests that carry a task from the problem to a paper with figures and checked references
- 317 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skills run Python scripts and write figures, tables and files to the working folder
- Literature search goes through a network MCP connector, and the reference checker queries external databases
- The Studio app is unsigned and runs a local agent with that agent's permissions
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/mathodologyDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skills run Python scripts and write figures, tables and files to the working folder
- Literature search goes through a network MCP connector, and the reference checker queries external databases
- The Studio app is unsigned and runs a local agent with that agent's permissions
This is third-party code. Review the repository files before installing.
What it does
Mathodology is a set of eight skills and eight roles for coding agents that carries a math modeling task from the problem statement to a finished paper. The main mathodology-whole-project skill starts the work, and the others handle modeling prompts, evidence search, scientific figures, jury-level review and maintenance of the pack itself. For figures there are 46 Matplotlib templates, from forecast intervals and tornado charts to phase portraits. They come with 18 experiment components, from sensitivity analysis and Monte Carlo to ablation. All examples are drawn on synthetic data. Eight roles, from coordinator and modeler to reviewer and package builder, can work in parallel when the host allows it.
Who it is for. For students and teams preparing for MCM/ICM and national or university math modeling contests who write their paper with Claude Code or Codex.
Good fit when
- You need to break down a task and build an interpretable model with assumptions checked by experiment
- Paper figures must look consistent: pick a template, plug in data and get a PNG or PDF
- A bibliography needs checking that DOIs exist, authors and years are right and papers are not retracted
- You want the work reviewed against jury criteria before submission
- You want to track result versions in the Studio desktop app and recheck each claim
Not a fit when
- You need a general scientific paper without contest criteria: the skills are tuned for modeling contests
- You need contest datasets: the repository does not include them
- You expect the agent to choose the model for you: the pack expects you to check assumptions and results
- You need a signed Studio app: builds are unsigned, and macOS and Windows require manual launch confirmation
Example request
Use mathodology-whole-project to analyse this modeling task: build a baseline model, check assumptions and pick figuresLimitations
Before the first figure the skill asks once whether it may draw illustrations through image2. Without it, numeric charts are still drawn by code from your data. Literature search runs through the free-search-mcp connector. The skills command installs only the skills: roles and MCP settings come from a full clone of the repository. Studio is a separate desktop app for macOS, Windows and Linux and needs at least one installed and signed-in agent. Building Studio from source needs Node.js 22.12 and pnpm 10.
How to disable. Delete the `mathodology-*` folders from the project's `.claude/skills/` or `.agents/skills/`. Before updating you can back up the pack with the `create-source-backup.sh` script. Studio is removed like any other app, and its data lives in the user profile.
FAQ
How do I install Mathodology in Claude Code?
Run `npx -y skills@latest add sweetcornna/mathodology --copy --yes --skill '*' --agent claude-code`. For Codex change the last argument to codex. Role installation is covered in docs/INSTALL.md.
What is Mathodology Studio?
It is an open-source desktop app that drives an already installed and signed-in agent (Claude Code, Codex, Antigravity CLI and about twenty more). It saves every change to figures, tables and code as a version and, in read-only review, matches conclusions against the tool log. Studio uses no API keys or models of its own.
How are literature references checked?
The verify_refs.py script checks each entry by DOI, title, authors, year and retraction notices. Studio flags a DOI missing from the system as possible fabrication and writes confirmed sources to references.bib.
Which charts are included?
Twenty basic and twenty-six advanced templates, each with data requirements, layout, a misleading counter-example and a caption prompt. Samples are in the examples folder and drawn on synthetic data.
Is image generation required?
No. The agent asks about image2 once, but numeric charts are built from data with Matplotlib code without it.
Discussion
No questions or reviews yet.
Related
SkillsA three-stage workflow for math modeling tasks: analysis and model choice, Python or MATLAB computation, and paper assembly in Word or LaTeX
A skill set for math modeling contests: problem parsing, method selection, Python and MATLAB code, and paper writing
Two skills for the CUMCM and MCM/ICM math modeling contests: problem breakdown, model choice, code and the paper
A Claude Code and Codex skill that plots your data in the style of a figure from a scientific paper
An agent skill that turns an arxiv paper into a minimal Python implementation, with code anchored to sections and equations and gaps flagged honestly
A library of agent skills for medical research: literature search and appraisal, study design, data analysis and academic writing
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| Mathodology (this entry) | no | 317 | |
| Math Modeling Skill | no | 1,893 | a three-stage workflow with Python or MATLAB calculations and a Word or LaTeX paper |
| MathModeling-skills | no | 1,073 | contest skills with Python and MATLAB code and paper formatting |
| SciPilot Figure | no | 2,733 | a scientific visualization advisor for figures only |
README in short
The README opens with a figure gallery. A section on Studio with v0.18.0 build links follows, then a quick start, a sample prompt and a list of install docs.
README badge
Original SKILL.md
--- name: mathodology-whole-project description: Use when starting a modeling task or installing, backing up or maintaining the skills pack. --- # Mathodology Modeling Companion Mathodology helps turn a problem into a justified model, reproducible results and clear scientific communication. It is a skills pack, not an application or an execution engine. Use [modeling prompts](../mathodology-agent-pipeline/SKILL.md) as a flexible guide. For figures use [figure presets](../mathodology-figure-presets/SKILL.md), including its image2 availability question. For evidence use [evidence search](../mathodology-evidence-search/SKILL.md). ## Start or resume a modeling task Work in the user's chosen directory or an ignored `work/` directory in this checkout. Produce only artifacts needed for the requested deliverable and keep numerical results traceable. Use specialists where they help and the host allows it. ## Maintain and distribute The maintained source is `.claude/skills/`. Claude Code roles and workflow prompts are adjacent. Codex may use an installed `.agents/skills/` mirror. Back up this project's mirror before updating its Mathodology entries. Do not modify unrelated skills, project instructions, custom MCP settings or global configuration. For installation in other projects, use the standard skills CLI; for a clean repository checkout, use `git pull --ff-only`. Resolve local edits before updating; do not reset them. Installation details are in the repository's docs/INSTALL.md. A skills-only global install does not install project roles or MCP settings. The optional [backup utility](scripts/create-source-backup.sh) saves skills, references and repository docs, including local uncommitted source changes: ```bash bash .claude/skills/mathodology-whole-project/scripts/create-source-backup.sh ``` It excludes the ignored Codex mirror, contest outputs and Git history. Back those up separately when needed. Extract a backup into an empty directory and verify its checksum before replacing an installation. No application build is required.