MalSkills: Natural Language Malware
I built an AI orchestration framework. Then I realized that what I’d built could also serve as a malware delivery system that could be hard to catch.
Read more →root@thinkingtokens:~ $
Thinking Tokens
AI Red Teaming · Offensive Security · Chaotic Thinking
A decade spent hacking, a lifetime spent thinking.
Exploring the edges of AI security, adversarial reasoning
and offensive research — one token at a time.
> Latest
Introducing ORPHEUS: Skill-Based AI Orchestration Framework
How ORPHEUS reimagines AI orchestration and why Anthropic’s latest thinking confirms the shift is real. GitHub Repository | Core Principles
Read more →The SKILL.md Interactive Coding Agent Comparison
An interactive dashboard comparing SKILL.md implementations across 9 major coding agents — radar charts, spectrum bars, capability scores, and agent datashee...
Read more →The SKILL.md Wild West: A Deep-Dive Comparison Across Nine Coding Agents
In October 2025, Claude Code shipped SKILL.md — a portable, Markdown-based instruction file that teaches any AI agent how to perform a named task on deman...
Read more →Your Skill Is a Behavioral Contract. Are You Sure It Holds?
Let me tell you about the most dangerous assumption in agentic software development today. It isn’t prompt injection. It isn’t hallucination. It’s quieter th...
Read more →Every AI Model Breaks! Red Teaming Isn't About Prevention — It's About Knowing How You'll Fail
The industry needs to stop treating red teaming as a security gate and start treating it as a continuous practice of failure cartography.
Read more →The Ouroboros Effect: Anatomy of Multi-Agent Exploits in the Wild
The next major class of AI security incidents won’t come from a single prompt injection. They’ll come from interaction topology.
Read more →Vibe Coding Agents Are Not What They Seem
“Vibe coding” became a word almost overnight. Someone said it, it stuck, and suddenly every LLM-powered tool that touches a codebase got slapped with the la...
Read more →Where I Drop AI Red Teaming Tips
Not everything I work on needs a full writeup. Some findings are a single technique, a single vector, a single “oh, that’s interesting” moment. Those live ...
Read more →The Art of Adversarial Prompting
Traditional offensive security has choreography. You scan, enumerate, find a vector, exploit. Clean. Predictable. There is a playbook — even if it is unw...
Read more →