Skip to main content

Agentic AI Security PathIncluded withMax plan.

Meet the next frontier: Agentic AI Security Path

Our purpose-built agentic platform puts real AI agents in your hands, each with its own tools and credentials. Hack them, defend them, and master the full OWASP Agentic Top 10.

Why take the Agentic AI Security Path?

We built you a live system to break. Attack it, watch the damage spread, then lock it down at your own pace. It’s agentic AI security you learn by doing, not from slides.

Start learning
  • Train where the agents fight back

    • Agents with their own models, identities and tools
    • Attacks that fire every time, so you practise until it sticks
  • Be the outlaw. Then be the law

    • Follow every tool call, message and identity in the inspector view
    • Red and blue skills from one path
  • Stop the attacks that already happened

    • Run the real attack patterns, hands-on
    • Build the control that would have stopped each one

Every bounty is a skill you keep

Ten modules, the full OWASP Agentic Top 10, and a skill you can use at work after each one.

TRANSMISSION // TEASER_01

Choose your trail

This path picks up where our AI Security Path leaves off. New to AI security? Start there. Already done it? Head straight out.

Step 01Prerequisite

AI Security Path

New to AI security? Learn the basics first.

Step 02This path

Agentic AI Path

Finished the AI Security Path? You're ready.

FAQ

A hands-on path that teaches you to attack and defend AI agents: systems that plan, call tools, hold credentials and talk to other agents. It covers the full OWASP Top 10 for Agentic Applications (ASI01 to ASI10) across five modules.
A multi-agent environment we built for this path. You talk to an orchestrator agent, which delegates to sub-agents that each run their own model, identity and tools and communicate through a shared message queue. The inspector view shows everything they did, so you can trace an attack and fix it.
The agents behave like normal, capable agents. The vulnerabilities are planted in their tools and data, so each attack works the same way every time you run it.
Security practitioners, AppSec and pentesters, SOC and incident responders, and engineers building agents who want to know how they fail.
It's the prerequisite. This path assumes you know prompt injection, RAG and data poisoning, LLM supply chain and secure AI architecture.
The full path is included with MAX. The first module is free for everyone.
No, this is a learning path. If you want to prove your AI security skills, see AI1.
Most are lecture-first or instructor-led. Here you work inside a live multi-agent system, on both sides, at your own pace, across all ten OWASP Agentic categories.

Included withMax

Agentic AI Security is part of MAX, our most powerful plan, built for practitioners who need to stay ahead. Try the first rooms with Premium, then upgrade for the full path and everything else MAX includes.

Upgrade to MAX

What you get with MAX:

  • Full Agentic AI Security Path: all modules and the agentic platform
  • Advanced paths: Red Teaming, SOC Level 2 and more
  • Cloud labs: attack and defend AWS and Azure
  • Recent Threats: hands-on labs for the latest vulnerabilities
  • Persistent AttackBox: your files and tools kept between sessions
  • 40% off a certification: one per year, on annual plans

The bounty's live

Ten outlaws on the OWASP Agentic Top 10. Ride in.