blog
Blog
Long-form writing on AI-assisted decompilation, reverse engineering, and what it takes to point fleets of coding agents at genuinely hard problems.
How agent swarms decompile games byte-for-byte
Seven months teaching fleets of coding agents to reconstruct a shipped game binary byte-for-byte — and building the oracles that keep them honest.
LLM-assisted decompilation: tools, not prompts
How fleets of coding agents byte-matched 18,000+ functions of a stripped Xbox 360 binary in eight weeks — by never letting the model judge its own work.
The Milo engine decompilation saga
How one engineer and a fleet of AI agents reverse-engineered a shared game engine across three games and two consoles — and discovered that a perfect match score can still be wrong.
Forking the decompilation toolchain
What it took to make nine open-source reverse-engineering tools — from Ghidra on down — fluent in a console ISA none of them were built to read.
From emulator to real console
Five months, three repositories, and one recurring lesson: the gap between an emulator and the real hardware is where all the interesting bugs live.
Infrastructure for running agent swarms
What it actually takes for one engineer to keep a fleet of AI coding agents fed, contained, and productive on hard problems.
Reverse engineering across domains
What happens when you point one AI-assisted decompilation workbench at firewall firmware, two pinball machines, a 2001 MMO client, and a home server — and let bytes, not vibes, decide who is right.