eslint-plugin-mushi-mushi

ESLint rules that catch the dominant agentic-coding failure modes: empty handlers (Gate 1) and mock-data leaks into production paths (Gate 2).

Rules

| Rule | What it catches | | ---- | ---- | | mushi-mushi/no-dead-handler | onClick={() => {}}, () => null placeholders, console.log-only handlers, throw new Error('not implemented') stubs, useCallback(() => {}, []). | | mushi-mushi/no-mock-leak | import { faker } from '@faker-js/faker' outside test paths, MSW imports, hardcoded John Doe arrays, lorem ipsum strings, [email protected]. |

Install

npm install --save-dev eslint-plugin-mushi-mushi

Use (flat config)

import mushi from 'eslint-plugin-mushi-mushi'

export default [
  mushi.configs.recommended,
]

Use (legacy)

{
  "plugins": ["mushi-mushi"],
  "extends": ["plugin:mushi-mushi/legacy"]
}

Why

Agentic / "vibe coding" tools scaffold UI fast — and ship handlers wired to nothing or pages padded with John Doe fake data. Sentry doesn't catch it (no error). User feedback catches it late (you already shipped). These two lint rules close the gap deterministically — no LLM, no false positives at zero cost.

Part of the Mushi Mushi v2 bidirectional bug knowledge graph. Works perfectly fine standalone.

Allowlist

Add // mushi-mushi-allowlist: <reason> immediately above a handler call site to opt out of no-dead-handler for that one occurrence. The reason is required — silent allowlists are the failure mode this plugin exists to catch.

License

MIT.