# Robustness Checker

Design and run risk-targeted robustness, sensitivity, error, and baseline checks for an...

- Repository: https://github.com/zhnnky329/MathModeling-skills (1073 stars)
- Category: Design
- License: MIT
- Last updated: 2026-09-24
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add zhnnky329/MathModeling-skills --skill robustness-checker -g -a claude-code -y`
Codex: `npx skills add zhnnky329/MathModeling-skills --skill robustness-checker -g -a codex -y`
Cursor: `npx skills add zhnnky329/MathModeling-skills --skill robustness-checker -g -a cursor -y`
Gemini CLI: `npx skills add zhnnky329/MathModeling-skills --skill robustness-checker -g -a gemini-cli -y`
Claude Code plugin marketplace: `/plugin marketplace add zhnnky329/MathModeling-skills`

## Description

Design and run risk-targeted robustness, sensitivity, error, and baseline checks for an approved mathematical model, emitting compact machine evidence in lean mode and a final report in submission mode.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Purpose

Test the claims most likely to fail. Choose checks from the model's assumptions and decision risks rather than filling a generic checklist.

# Preconditions

- Approved main and usable baseline executed.
- Run summary, method card, probe summary, and relevant outputs exist.
- Claim or decision to be tested is known.

# Workflow

1. Identify load-bearing assumptions and claims.
2. Select applicable checks:
   - parameter or weight perturbation;
   - alternate split or resampling;
   - seed stability;
   - outlier/missing-data treatment;
   - constraint/capacity perturbation;
   - baseline comparison;
   - output concentration/rank stability;
   - error and uncertainty analysis.
3. State perturbation ranges and why they are meaningful before interpreting results.
4. Run checks with fixed seeds where stochastic.
5. Save compact metrics to:

`robustness/Qx/qx_robustness_summary.json`

6. In `submission`, also save:

`robustness/Qx/qx_robustness_report.md`

7. If the stability verdict affects method continuation or claim scope, invoke one choice card and log the human answer in `qx_decisions.jsonl`.

# Summary Contract

Record:

- tested claim/assumption;
- input and result source paths;
- perturbation;
- metric and threshold if predeclared;
- observed value;
- status `PASS`, `CONDITIONAL`, or `FAIL`;
- limitation;
- fallback-trigger relevance.

# Rules

- Do not run irrelevant checks merely to reach a count.
- Do not invent a threshold after seeing the result without labeling it exploratory.
- Do not convert stability metrics into the human confidence verdict.
- Do not create `robustness-checker_modeler_decision.md`.
- A failed robustness check is evidence for adjust/fallback/claim downgrade, not permission for AI to decide.

# Verification

The rest: https://github.com/zhnnky329/MathModeling-skills/blob/HEAD/.claude/skills/robustness-checker/SKILL.md