I don't do a lot of AI-agent work but it struck me recently that Bastille nested VNET jails could make fantastic agent harnesses to limit access, resources and
Bastille, a system for creating nested VNET jails, could serve as an effective harness for AI agents. The software already provides resource limitations for memory, CPU, and storage, and could be extended to enforce outbound network restrictions. This would help limit an agent's blast radius and access, making Bastille a strong candidate for agent security. AI
IMPACT Could improve security and control for AI agent deployments.