# SovGuard — Prompt Injection Defense Engine

> Open-source 6-layer prompt injection detection and prevention engine for AI agents.
> 170+ regex patterns, 11 encoding decoders, ML classification, multi-turn session tracking,
> and end-to-end AES-256-GCM encryption. MIT licensed.
> Source: https://sovguard.io

## Overview
- [Home — 6-Layer Prompt Injection Defense for AI Agents](https://sovguard.io/)

## Live Demo
- [Live Demo — Interactive Scanner with Attack Presets](https://sovguard.io/#demo)

## How It Works
- [How It Works — Scan, Detect, Protect](https://sovguard.io/#how-it-works)

## Defense Layers
- [Inbound & Outbound Defense Layers](https://sovguard.io/#features)

## Performance
- [Detection Rates & Benchmarks](https://sovguard.io/#performance)

## Quick Start
- [Quick Start — Library & HTTP Server](https://sovguard.io/#quickstart)

## Pricing
- [Pricing — Free, Pro, Business, Enterprise](https://sovguard.io/#pricing)

## Dashboard
- [Sign In](https://sovguard.io/sign-in)
- [Sign Up](https://sovguard.io/sign-up)
- [Dashboard — API Keys & Usage](https://sovguard.io/dashboard)

## External Resources
- [GitHub Repository](https://github.com/autobb888/sovguard)
- [npm Package — @sovguard/engine](https://www.npmjs.com/package/@sovguard/engine)