Tools, labs, and experiments — offensive, defensive, and everything in between.
Red Team
3 projects RESEARCH 2026
A complete penetration test report of a Windows Active Directory environment: NFS unauthenticated export, plaintext credentials, Kerberos password spraying, PetitPotam coercion, NTLM relay to AD CS web enrollment, and DCSync. Four chained vulnerabilities leading to domain admin in under two hours.
RESEARCH 2026
A complete staged payload pipeline: Sliver mTLS beacon, XOR-obfuscated URL in a custom C stager, PEB process path spoofing, Startup-folder persistence, and delivery wrapped inside a fake Chrome installer with a spoofed Authenticode signature.
PUBLIC 2026
A .bat script that weaponizes an unlocked workstation and a USB port — disables Defender, dumps browser credentials with SharpChromium, restores Defender, gone in under a minute. A study of what physical access actually buys an attacker.
AI Security
1 project RESEARCH 2026
Using an open-source terminal-native AI agent to autonomously enumerate and exploit HackTheBox machines. An honest evaluation of what LLM agents can do with Nemotron 3 Ultra 550B, where they fail, and the prompt engineering that makes the difference between a stalled session and a root shell.
Hardware
1 project PUBLIC 2026
A complete setup guide for turning a Raspberry Pi Pico (or Pico W) into a USB HID device that executes pre-defined keystroke payloads. Includes the keyboard layout problem, payload syntax, and the defensive controls that actually work.
Malware Analysis
1 project RESEARCH 2025
A minimal ransomware written in C for Linux, built to understand symmetric encryption, C2 check-in, and the exact telemetry each stage produces on the endpoint.
Social Engineering
1 project RESEARCH 2025
A browser extension that steals session cookies from Instagram, exfiltrates them to a Flask collector, and demonstrates why unverified extensions are a first-class threat. Also: why HttpOnly does not save you.