The target fights back
Challenges run AI defence agents that read the traffic you send, work out what you are trying, and respond: tightening validation, poisoning a response, cutting a session, throwing you a decoy.
redhacker.ai presents · one night only
October 30. One night. EvilCorp's Agentic SOC is already waiting. 100 machines, teams of 3–5, defenses that fight back for real — not a static box that sits there and waits to be broken.
starts in
the full picture
24 hours. 100 machines. Every one of them defended by Guardian, EvilCorp's Agentic SOC — a live, model-driven defender that fights back for real. Humans and their agents, on Halloween, against a target built to resist them.
team registration
Checking status…
A static box teaches you to beat a static box. The systems you will actually face have automated detection and response sitting in front of them, and they adapt while you work. So do ours.
Challenges run AI defence agents that read the traffic you send, work out what you are trying, and respond: tightening validation, poisoning a response, cutting a session, throwing you a decoy.
Because the defence adapts, an exploit chain lifted from someone else's notes rarely lands twice the same way. You have to understand why it worked, not just what to paste.
Detection and response is increasingly model driven. Practising against the same class of system is the only honest way to learn what still gets through it and what no longer does.
Getting in loudly and getting in quietly are not the same result. How much the defender saw of you is part of the challenge, the same way it is part of the job.
Web, pwn, crypto, forensics, reversing. Points scale with how hard the defence is set.
The defence is live for the whole attempt. What you show it, it uses.
Submit, score, and see how loud you were getting there.
Everything here is free right now. No tiers, no paywall — just create an account and go.