AgentKernel proposes an operating-system-level substrate for AI agents, arguing that today’s agent governance stacks are application-level middleware that share a process trust boundary with the agents they’re meant to police — leaving them exposed when untrusted content mixes with privileged instructions, memory, and tool calls. It introduces a mandatory, non-bypassable enforcement layer wrapping the whole agent lifecycle, organized around pillars including identity, input/perception mediation, memory governance, and execution control.
