🎯 What You'll Learn
- Understand how LLM apps build prompts
- Perform direct prompt injection
- Exploit indirect injection via content
- Extract data and override instructions
Overview
Prompt injection is the defining vulnerability class of LLM applications. This lab covers direct and indirect injection, instruction override and data exfiltration from AI-powered apps.
Core Topics
- Prompt construction
- Direct injection
- Indirect/stored injection
- Mitigation patterns
Prerequisites
A working KaliRange lab environment and comfort with the Linux command line.
Recommended Workflow
- Spin up the target in your KaliRange lab environment and confirm connectivity.
- Enumerate the target thoroughly before touching any exploit — information first.
- Reproduce each technique by hand so you understand why it works, not just the command.
- Capture evidence (commands, output, screenshots) as you go.
- Write a short note on how a defender would detect or prevent what you just did.
Only ever run these techniques against systems you own or have explicit written permission to test. Practise inside your own KaliRange lab.
Your Goal
Work through every task in your own lab, document your findings as you would on a real engagement, then note the defensive takeaways.
Ready to practise. Work through the steps above at your own pace, then move on to a related lab.