Jason Jeske
Agentic AI Engineer, Offensive Security
I am an agentic AI engineer working in offensive security: building AI agents that probe the LLMs, AI agents, MCP servers and RAG pipelines organisations are shipping faster than they can secure them.
Two decades in offensive security before the agentic era, across enterprise and consultancy. I started as a teenager, inspired by the late Kevin Mitnick, and never stopped.
Krypteia Sec is where that work gets published. Research and development in the open, covering prompt injection and jailbreaking, MCP and tool-layer security, agent kill chains, and the detection engineering that catches them.
Long-form research and field guides are agent-drafted and verified by a human before they ship, which is stated plainly here because readers deserve to know how what they are reading was produced.
Krypteia Sec is an independent research and development platform, run in a personal capacity. The research published here is my own and does not represent, and is not funded or endorsed by, any employer or client.