> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nullboard.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Get started with the @impulsedev/ink library.

<div className="flex justify-center items-center w-full">
  <img className="block dark:hidden" src="https://mintcdn.com/nulltools-7f103e0d/rWxnnTMEBQlCXIOV/logo/ink.png?fit=max&auto=format&n=rWxnnTMEBQlCXIOV&q=85&s=95e73af477d568a2183b762aca3659da" alt="Ink Logo" width="400" data-path="logo/ink.png" />

  <img className="hidden dark:block" src="https://mintcdn.com/nulltools-7f103e0d/rWxnnTMEBQlCXIOV/logo/ink.png?fit=max&auto=format&n=rWxnnTMEBQlCXIOV&q=85&s=95e73af477d568a2183b762aca3659da" alt="Ink Logo" width="400" data-path="logo/ink.png" />
</div>

**@impulsedev/ink** is a specialized library designed to recognize handwritten digits and mathematical operators directly in the browser. It features a unique **client-side auto-training** capability, allowing it to adapt and learn without heavy server-side models.

## Key Features

* **Math Recognition**: accurate detection of digits (0-9) and operators (+, -, \*, /, =).
* **Auto-Training**: The library can train its own lightweight CNN model in the browser using bundled sample data. Zero server configuration required.
* **Expression Evaluation**: Automatically parses recognized characters into valid mathematical expressions and calculates the result.
* **Stroke Management**: Built-in tools for capturing, storing, and manipulating handwriting strokes.
* **TypeScript Support**: Fully typed for a robust development experience.

## Installation

Install the package via npm:

```bash theme={null}
npm install @impulsedev/ink
```

## Quick Start

Here is a minimal example to initialize the library and recognize a simple drawing.

```typescript theme={null}
import { Ink } from '@impulsedev/ink';

// Initialize Ink
// By default, this starts the auto-training process in the background
const ink = new Ink();

// Wait for the model to be ready (trains or loads)
await ink.loadModel();

// ... (Integration with your canvas to capture strokes) ...

// Recognize the current drawing
const result = await ink.recognize();

if (result.valid) {
  console.log(`Expression: ${result.expression}`); // Output: "2 + 2"
  console.log(`Result: ${result.result}`);         // Output: 4
}
```