Prompt Injection Defense

---
name: prompt-injection-defense
version: 1.0.0
description: "Four-layer prompt injection defense for AI chatbots on websites — client-side input detection, system prompt hardening, output filtering, Nginx-level auth injection"
author: agent
tags:
  - security
  - prompt-injection
  - sentry
  - guardrails
  - website-security
  - ai-chatbot
platforms: [windows, linux, macos]
---

# Prompt Injection Defense

Multi-layer defense architecture for AI chatbots embedded in websites. Prevents attackers from manipulating the AI's system prompt, extracting sensitive data, or bypassing authentication.

**When to load**: user asks about prompt injection defense, sentry mechanism, AI chatbot security, guardrail implementation, or security hardening of an AI widget on a website.

---

## Architecture: Four-Layer Defense