rainmanjam/poka-yoke
22 stars · Last commit 2026-09-01
Mistake-proofing for software. Shigeo Shingo's poka-yoke method as a Claude Code plugin: audit code for footguns, design APIs that resist misuse, install guardrails, turn incidents into devices.
README preview
<p align="center">
<img src="docs/assets/banner/readme-header.png" width="100%"
alt="poka-yoke: an andon signal tower beside the wordmark, with the severity ladder: green Control, the mistake is impossible; amber Warning, announced as it happens; red Detection, found afterwards.">
</p>
# Poka-Yoke
> **An agent will tell you what to fix. It will rarely tell you what its fix makes impossible.**
>
> Unprompted, models close a design by naming what it forecloses **45%** of the time. With
> these skills, **80%**, measured across 132 graded verdicts and six model families. That one
> habit is most of what this does, and it is the difference between advice you agree with and
> a constraint you can rely on.
>
> The average conceals where the effect lives. Asked squarely to design an interface, models
> already do this 77% of the time unprompted and the skills add eleven points. The large gains
> are in the scenarios where nobody asked for a design review — see the table below.
>
> A dependency-free **hazard scanner** for TypeScript, Python, Go, Rust and SQL; installable
> **pre-commit / CI / lint / hook devices**; and **11 skills** that apply the same method while