ABOUT / KARMAZY LABS

A small lab. A serious question.

How do we secure software that is increasingly built by AI, connected to AI, and operated by AI?

K4RM4ZY L4B
KARMAZY LABS

Security for the AI-native world.

Karmazy Labs is an early-stage AI security research and engineering project focused on understanding and securing AI-native systems.

Our work connects adversarial testing, practical security tooling, and careful research. VibeGuard is our first product direction, with AgentGuard as the next research and product priority.

We’re building toward a long-term security technology company. For now, that means keeping the scope focused and being clear about what is planned, what is in progress, and what has evidence behind it.

What we’re interested in.

Prompt InjectionIndirect Prompt InjectionAgent Tool AbuseAI-generated Code SecuritySystem Prompt SecurityRAG SecuritySecurity toolingAdversarial AI testing

OUR APPROACH

Ambition, without invented certainty.

Build sequentially. Study failure. Explain the findings. Share limitations alongside progress.