@mima-ai/eslint-plugin-mima

ESLint rules for detecting unattested AI SDK calls in TypeScript and JavaScript.

Catches AI library calls that are not wrapped with mima.wrap() at lint time — in your editor and in CI, before they reach production.

Install

npm install --save-dev @mima-ai/eslint-plugin-mima

> **Migrating from `eslint-plugin-mima`?** Replace the package name and update plugin references from `"mima"` to `"@mima-ai/mima"` in `.eslintrc`, or keep using the alias `mima` in flat config.

Configure

ESLint flat config (ESLint 8+)

// eslint.config.js
import mima from '@mima-ai/eslint-plugin-mima';

export default [
  {
    plugins: { mima },
    rules: {
      'mima/no-unattested-ai-call': 'error',
    },
  },
];

Legacy .eslintrc

{
  "plugins": ["@mima-ai/mima"],
  "rules": {
    "@mima-ai/mima/no-unattested-ai-call": "error"
  }
}

Or use the built-in configs:

// eslint.config.js
import mima from '@mima-ai/eslint-plugin-mima';

export default [
  mima.configs.recommended,  // warn
  // or:
  mima.configs.strict,       // error
];

What it flags

// Error — unattested OpenAI call
const response = await openai.chat.completions.create({ ... });

// Error — unattested Anthropic call
const msg = await anthropic.messages.create({ ... });

// OK — wrapped with mima
const generate = mima.wrap('generate', async (prompt: string) => {
  return await openai.chat.completions.create({ ... });
});

AI libraries covered

| Library | Methods flagged | |---|---| | openai | chat.completions.create, completions.create | | @anthropic-ai/sdk | messages.create, messages.stream | | @google/generative-ai | generateContent, generateContentStream | | @azure/openai | getChatCompletions, streamChatCompletions | | ai (Vercel AI SDK) | generateText, streamText, generateObject, streamObject | | langchain / @langchain/core | invoke, stream, batch | | cohere-ai | chat, generate, embed | | @mistralai/mistralai | chat, chatStream |

CI integration

# .github/workflows/lint.yml
- name: Lint
  run: npx eslint src --ext .ts,.tsx

Non-zero exit on violations blocks the merge.

Inline disable

For intentional unattested calls (health-check pings, etc.):

// eslint-disable-next-line mima/no-unattested-ai-call
const ping = await openai.models.list();

Relationship to other Mima tools

| Layer | When | What it catches | |---|---|---| | ESLint plugin | Lint time / editor | Statically detectable unattested calls | | mima scan | CLI / CI | All AI call sites via AST, any library | | mima.wrap() | Runtime | Forces wrapping at the call site |

The ESLint plugin is the earliest signal. Use all three for defence-in-depth.

Links

• Full ESLint plugin docs
• TypeScript SDK
• mima.works