mpm
// Michael McCord

I break protocols and
build the agents that watch them.

Security architecture, protocol-level vulnerability research, and the governance scaffolding that keeps autonomous offensive tooling honest.

scroll
Selected work
01

AttackBench

An autonomous pentester that can't talk itself into firing the gun.
02

Breaking Tor on a Budget

CVE-2014-5117. We unmasked the network for less than a used car.
03

Multi-agent Orchestration

Why one agent shouldn't both find the bug and decide to exploit it.

I take on high-consequence problems — the kind where the failure mode is on the news, not in the backlog. Forensic engineering at CMU/CERT and the Citi Fusion Center; protocol research that earned a CVE; agentic systems built to act, not just suggest.