A lab for the agentic frontier.

MASec Lab is independent security research into multi-agent and embodied AI systems — the agents, the protocols they speak, and the robots and drone swarms they increasingly control.

The work runs on a simple conviction: the security industry is racing to test AI models, while the riskier surface — how agents coordinate, delegate, and act together — goes largely unexamined. MASec Lab exists to build the methodology and the tooling for that layer, and to publish it openly.

It is founded and run by evkir, a security researcher working in offensive security and agentic-AI safety.