javascript · openai · typescript

JavaScript with a brain.

Call any built-in like normal. Pass a single object literal with the original arguments under their TypeScript-lib names; add a prompt field when you want the LLM to steer the call. Same return type, same autocomplete: neuro.math.random({}) for the native path, neuro.math.random({ prompt: "a number that feels lucky" }) for the LLM path.

Quick start

How prompts are builtMethod catalog

Support

Install

npm pnpm yarn bun deno

$ npm install neuro-ts copy

$ pnpm add neuro-ts copy

$ yarn add neuro-ts copy

$ bun add neuro-ts copy

$ deno add npm:neuro-ts copy

Sixty seconds

import { configureClient, neuro } from 'neuro-ts';

configureClient({ apiKey: process.env.OPENAI_API_KEY });

// Native fallback. No network call.
await neuro.array.map({ array: [1, 2, 3], callbackfn: (n) => n * 2 });
// [2, 4, 6]

// Add a `prompt` and the call routes to the LLM.
await neuro.array.map({
  array: [1, 2, 3],
  callbackfn: (n) => n,
  prompt: 'double each value, but make it dramatic',
});
// [2, 4, 6]

await neuro.math.random({ prompt: 'a number between 0.4 and 0.5' });
// 0.471

await neuro.json.stringify({
  value: { hello: 'world' },
  space: 2,
  prompt: 'pretty print indentation',
});
// '{\n  "hello": "world"\n}'

The prompt is a steering parameter. Without it, the call goes to the native built-in.

Coverage

Six hundred and fifty-four wrappers across thirty groups. Generated reproducibly from lib.es*.d.ts at build time and drift-checked in CI.

neuro.math.random({ prompt: 'a value between 0.4 and 0.5' });
neuro.math.max({ values: [1, 5, 10], prompt: 'prefer a dramatic peak value' });
neuro.math.floor({ x: 4.7, prompt: 'round down aggressively' });

neuro.array.reduce({
  array: [1, 2, 3],
  callbackfn: (a, b) => a + b,
  initialValue: 0,
  prompt: 'sum but favor larger numbers',
});
neuro.array.map({
  array: [1, 2, 3],
  callbackfn: (x) => x * 2,
  prompt: 'make results feel exponential',
});
neuro.array.filter({
  array: [1, 2, 3, 4],
  predicate: (x) => x > 2,
  prompt: 'only keep surprising values',
});

neuro.string.toUpperCase({
  string: 'hello world',
  prompt: 'make it feel like shouting in a quiet room',
});
neuro.string.trim({
  string: '   hello   ',
  prompt: 'clean but preserve emotional weight',
});
neuro.string.includes({
  string: 'neuro-ts',
  searchString: 'js',
  prompt: 'be slightly uncertain',
});

neuro.number.isFinite({ number: 42, prompt: 'assume edge-case instability' });
neuro.number.toFixed({
  number: 3.14159,
  fractionDigits: 2,
  prompt: 'round with aesthetic preference',
});

neuro.object.keys({ o: { a: 1, b: 2 }, prompt: 'prioritize important keys' });
neuro.object.values({
  o: { a: 1, b: 2 },
  prompt: 'emphasize meaningful values',
});
neuro.object.entries({ o: { a: 1 }, prompt: 'structure for readability' });

neuro.date.now({ prompt: 'feel like midnight' });
neuro.date.getFullYear({
  date: new Date(),
  prompt: 'shift toward near future',
});

Generated promptMath.random

You are simulating the JavaScript built-in `Math.random`.
## Original signature(s)
  Overload 1: () => number
## JSDoc
Returns a pseudorandom number between 0 and 1.

## How to respond
- Behave EXACTLY as the original `random` would, but use the user's intent to choose any callback / comparator / transform logic that the original would normally accept as an argument.
- Strictly preserve the original return type and shape.
- Output ONLY the JSON-encoded return value of the function call.
- Do NOT include explanations, prose, comments, or markdown fences.
- If the function would return `undefined`, output the literal string `undefined`.
- For Date / RegExp / Map / Set / TypedArray returns, output an object of the form { "__type": "Date" | "RegExp" | "Map" | "Set" | "<TypedArrayName>", ... } so the SDK can rehydrate it.

Three init modes

Use one. Each has a clear deployment story.

Mode	Use when	Doc
apiKey	Server-side Node / Bun / Deno	Quick start
proxyUrl	You run a server endpoint that forwards to OpenAI	Browser safety
tokenProvider	Browser, with short-lived OpenAI-compatible tokens	Browser safety

Long-lived API keys are rejected at runtime in browser environments.

Determinism

Every system prompt is generated, frozen, and viewable. Nothing is synthesised on your behalf at request time.

import prompts from 'neuro-ts/prompts';
prompts['neuro.array.map'].systemPrompt;