# Abuse Detector (VOIDSEC)

**A powerful, plug-and-play abuse/toxicity detection module for Node.js bots and APIs.**

Built with a hybrid detection system combining lightning-fast rule-based matching and advanced AI to catch toxic, abusive, and harmful content effectively.

---

## ✨ Features

- ⚡ **Ultra-fast keyword detection** using Aho-Corasick algorithm
- 🧠 **AI-Powered Toxicity Detection** with TensorFlow.js
- 🔍 **Fuzzy Matching** to catch typos and intentional misspellings
- 🗣️ **Phonetic Normalization** – resistant to common bypass techniques
- 📚 **Self-Learning System** – automatically learns new abusive patterns
- 🔌 **Easy Integration** – perfect for Discord bots, Telegram bots, APIs, and more
- 📊 **Detailed Results** with match type and confidence

---

## 📦 Installation

```bash
npm install @sabir7718/abuse-detector

🚀 Quick Start

const abuse = require("@sabir7718/abuse-detector");

(async () => {
  const result = await abuse.check("You are so stupid idiot!");
  console.log(result);
})();

Example Output

{
  "abusive": true,
  "type": "rule_match",
  "category": "insult",
  "matches": ["stupid", "idiot"],
  "score": 0.95,
  "fuzzy": null
}

🤖 AI Detection Example

const result = await abuse.check("I hate you and want to kill you");
console.log(result);

Output:

{
  "abusive": true,
  "type": "toxicity",
  "score": 0.87,
  "category": "threat"
}

📚 Learning System

The module can automatically learn from new abusive content.

// Get pending entries for review
const pending = abuse.getPending();
console.log(pending);

// Approve or reject learned entries
abuse.approve(0);   // Approve index 0
abuse.reject(1);    // Reject index 1

📁 Required Files

Create the following directories and files in your project root:

/database/
  ├── abuse.json
  └── learning.json

Example abuse.json structure:

{
  "insult": [],
  "profanity": [],
  "sexual": [],
  "threat": [],
  "racism": []
}

⚙️ How It Works

1. Text Normalization – cleans and standardizes input
2. Aho-Corasick Search – fast keyword matching
3. Phonetic Analysis – detects bypass attempts
4. Fuzzy Matching – Levenshtein distance for typos
5. TensorFlow.js AI Model – fallback for complex cases
6. Self-Learning – stores and learns new patterns

📋 Requirements

• Node.js 16 or higher
• TensorFlow.js (@tensorflow/tfjs-node recommended)

📜 License

© 2026 SeXyxeon (VOIDSEC)

All rights reserved.
This project is protected under copyright.

Allowed:

• Personal and commercial use with proper credit

Not Allowed:

• Redistribution
• Claiming as your own work
• Selling without permission

🔗 Connect With Me

• YouTube: https://youtube.com/@voidsec7718
• Instagram: sabir._7718
• Telegram: https://t.me/SABIR7718
• GitHub: https://github.com/SABIR7718
• WhatsApp: +91 73650 85213

⚠️ Disclaimer

This module is intended for moderation and content filtering purposes only. The author is not responsible for any misuse of this software.

Made with ❤️ for the developer community