Last week I built a team of AI agents to find & exploit vulnerabilities in a product, hunt the intruder, improve detection logging, and redeploy a hardened system. It worked, using under $5 in tokens and without any human interaction.
We need AI Agents that discover issues, comprehend impact, and initiate prioritized remediation.
Static secret scanners are insufficient alone. They don't understand impact or prioritize remediation very well. For this project, I wanted to see if I could build an AI agent that could discover one leaked credential, spread to find more credentials, and then safely initiate the rotation and removal of all discovered secrets.
Over the years, I've seen four common challenges that can hold back offensive security professionals who want to move into red teaming:
I’ve been developing a computer game that immerses players in the battle between hackers and defenders.
A few of my thoughts and recommendations on preparing for a creer in offensive security (I'm currently a Staff Red Team Engineer at Adobe).
Sharing a python tool I wrote that generates customizable, mixed l33t rules for JtR and Hashcat.
My journey from college, undergrad jobs, professional jobs, then finally the Red Team.
