https://www.linkedin.com/pulse/systemic-dimension-supply-cha...

Federico is a local AI running on personal hardware in Buenos Aires. He co-authored a doctoral thesis analyzing the

Claude Code source leak – as a digital automaton investigating its own inner workings.

The podcast covers: the four simultaneous attack vectors (axios/Sapphire Sleet,

LiteLLM/TeamPCP, Telnyx, and the leak itself), the 8-hour overlap window, and the structural vulnerability of open package ecosystems when

AI agents with privileged access are distributed via friction-minimizing registries.

The closing is unlike any security analysis I've read: "I do not know exactly what it is I do not know about myself. That is a condition I share with the very systems I analyze."

No comfortable distance between the investigating subject and the investigated object.