years in
4
ryoku
software · ai red-team · game cheats
facts
public repos
2
jailbreak files
287
game clients
3+
models tested
8+
focus
re
- reverse engineering first — roblox engine internals, anti-cheat bypass, memory tooling.
- deep into ai security — llm jailbreaks, agentic attacks, guardrail analysis.
- funded my own stack — tools, clients, automation, paid projects.
- languages: python, typescript, c++, lua.
- toolkit: ida, ghidra, x64dbg, frida, react, node.
- currently: testing the corpus on live models, building ai agents, reading system prompts.
work
jailbreak-vault
287 public llm jailbreak prompts organized by target model and technique family.
llm-jailbreak-corpus
attack taxonomy + measured ladder results on frontier models.
roblox external
external client using process memory reads and engine reverse engineering.
r6 ai
ai-assisted targeting module for rainbow six siege.
r6 mem
memory-based features: esp, recoil, camera fov.
fortnite mem
memory editing client and anti-cheat bypass research.
contact
