Project Lunchbox Gives Every AI Agent a Kernel, a Freeze and a Fork

Edera wants AI agent execution to stay on your own Kubernetes clusters, and it shipped three primitives to argue the case.

• Each agent runs in its own microVM with a private Linux kernel, so network reach is cut below the guest.
• Live snapshotting freezes a running agent with its state intact, then restores it where it stopped.
• Copy-on-write forking clones a warm agent into isolated children that share memory until a write.
• Kubernetes 1.37 already ships pod checkpoint and restore as alpha, so the platform is growing the same idea from below.

The cost case matters as much as the security one. A fleet of thousands of agents holds memory while it waits on a tool call or a human review. Read the full analysis.