Sage Protects Your AI Agent From Prompt Injection

Your AI coding agent has full access to your terminal and file system. Sage is an open source Agent Detection & Response (ADR) layer from Gen Digital (Norton/Avast) that sits between your AI agent and your OS, intercepting every tool call in real time.

In this video, we take a deep look at how Sage works under the hood — its four detection layers (local heuristics, URL reputation, package supply chain checks, and plugin scanning), the elegant data flow architecture, and how to set it up on Claude Code, Cursor, VS Code, OpenClaw, and OpenCode.

0:00 Intro
0:42 What is Sage (ADR)?
1:43 Supported Platforms
2:22 Detection Layers Overview
2:28 Layer 1: Local Heuristics
4:58 Layer 2: URL Reputation
5:41 Layer 3: Package Supply Chain
7:06 Layer 4: Plugin Scanning
8:01 Data Flow Architecture
11:15 Codebase Architecture
13:03 Trusted Domain System
14:04 Getting Started & Installation
18:14 Should You Use Sage?

Sage on GitHub: https://github.com/gendigitalinc/sage

Видео Sage Protects Your AI Agent From Prompt Injection канала KTG Analysis