I was hunting on a private YesWeHack program and I kept hitting the same wall. Not a hardened application, a 403: 461 bytes of HTML, server: AkamaiGHost, back in forty milliseconds. Whenever I got close to something the edge ate the request before the origin saw it. Three separate leads died that way across two sessions, and after the third one I stopped treating it as bad luck.

Here is a service principal name: cifs/DC1 And here is another one: cifs/DC1‌ They are not the same string. The second one ends with U+200C, a zero width non-joiner. Three bytes, e2 80 8c, that your terminal, your browser and your eyes all agree to pretend do not exist.

I wanted to get better at reading code for security issues - not just knowing vulnerability names, but being able to sit down in front of an unfamiliar codebase and find things systematically. This is what I tried, what worked, and the mental model I ended up building around taint analysis. Python-first, but the approach transfers.

A few days before leHack 2026 I gave a talk at the HackTheBox Meetup on AI pipelines for bug bounty. Here is the whole deck, plus a detailed walkthrough of every technical slide with code, sourced numbers, and links to the research behind each idea.

I want an AI agent that can do offensive and defensive security work without ever leaking a credential, a hostname, an IP or a domain to the model provider, and to keep that property no matter which provider sits behind the API. This is Part I of the research. It covers the threat model, the state of the art, the core mechanism (bidirectional tokenization with host-side resolution), and four experiments that run on real HackTheBox machines, including an autonomous agent that drives a real domain controller while seeing nothing but opaque tokens.

In Part I, I caught secrets by knowing the shape of secretsdump, of netexec, of hashcat. A reader could fairly ask: are you doing research, or are you writing regexes forever? They would be right. This part inverts the whole thing. You cannot enumerate every secret format in the world, and some secrets have no format at all, so stop trying. Keep only what is provably generic (dictionary words, numbers, dates, protocol constants) and tokenize the rest by default. I measure it live on HTB Shibuya through a full root chain, then across ten Active Directory machines, with no per-tool rules at all, and I situate it against the literature, which turns out to have a clear closest neighbor and a clear gap.