Decepticon: Open-Source Multi-Agent AI for Autonomous Pentesting

😊 This time, Twitter has been hyping Claude Code — though only for autonomous end-to-end pentesting across the entire attack chain.

It’s called Decepticon — a system of 5 AI agents named in a Transformers style.

You set a target IP, and then the platform, inside an isolated Kali environment, runs the whole scenario on its own: reconnaissance, vulnerability discovery, exploitation, persistence, credential harvesting, and lateral movement.

With no human in the loop. Open source.