Containers aren't tiny fortresses. They're leaky rowboats unless you know what you're doing. This hands-on workshop demystifies container security layer by layer, showing how real-world missteps in runtime, image, and host configurations open doors to escapes, persistence, and lateral movement. We start by dissecting how containers actually work, walk through common isolation failures, and demonstrate how attackers exploit weak assumptions. Next, we'll consider how this threat changes based on the type of workload (server, AI, CI/CD, etc), if you're on a server or desktop, and how a runtime can change everything - common "isolation tricks" aren't very isolating at all! Then we'll explore countermeasures for escape, such as eBPF runtime protection tools, mandatory access control policies (such as AppArmor or SELinux), syscall filtering via `seccomp`, and microVM runtimes. Whether you're building, securing, or regulating containerized apps, attendees will leave with a threat model, practical tools, and maybe a new trick or two for _literally_ popping out of the box.

Participants should have a laptop with internet connectivity and be comfortable at the command line. If they want to stay local, they'll need a virtualization solution like VirtualBox (https://virtualbox.org) or UTM (https://mac.getutm.app). Otherwise they'll need the ability to spin up a Linux VM on the cloud of their choice (AWS, Digital Ocean, etc.). I'll provide a shell script to handle installing dependencies for a vanilla Ubuntu base. If you want, I can grab a Raspberry Pi to set up a local network share with a pre-built VM image too.