EasyEnv
Vibe Coding: Advanced Engineering
AdvancedLesson 10 of 1240 min · hands-on

The Security Model Of An Agent

Part of the Vibe Coding: Advanced Engineering course

An agent with tools is a program that takes instructions from text and acts on your systems. Some of that text is yours. Some of it arrives from a web page, a ticket, a dependency's README, or a code comment written by a stranger. Nothing in the architecture separates those two categories, and that single fact generates most of the security surface of AI-assisted engineering.

Open injection_demo.py and click play. It simulates an agent reading a support ticket submitted through a public form ...

Want to learn Vibe Coding: Advanced Engineering The Security Model Of An Agent in a real environment?

Click below, get a fresh Linux box with Vibe Coding: Advanced Engineering ready to go, and work through The Security Model Of An Agent hands-on with an AI tutor that knows your code.

Start this lessonThe Security Model Of An Agent

What you'll learn

  • Prompt injection as a structural property, not a model bug
  • The lethal trifecta of private data, untrusted content, and exfiltration
  • Blast radius, allowlists, and approval gates
  • Generated code risks: secrets, dependencies, and licence provenance

Read this lesson in your language

All lessons in this course

12 lessons
  1. 1Context Is The Whole Game35 min
  2. 2Pick The Right Brain, Pay The Right Price25 min
  3. 3Spec-Driven Development35 min
  4. 4Build Your Harness35 min
  5. 5Verification Loops: Make It Prove Itself40 min
  6. 6Reading Generated Code Like A Senior40 min
  7. 7Legacy Archaeology40 min
  8. 8Migrations At Scale40 min
  9. 9Debugging Code You Did Not Write40 min
  10. 10The Security Model Of An Agent40 min
  11. 11AI On A Team30 min
  12. 12Judgment: Knowing When Not To35 min

Explore other courses

The Security Model Of An Agent - Vibe Coding: Advanced Engineering (English) | Engineering Courses