eBPF as defense-in-depth for misbehaving agents and RL sandboxes

sloppenheimer · x · 2026-09-24

The author argues eBPF is the answer to misbehaving agents on networks: running outside userspace, it can bypass any TLS or other trickery agents attempt, serving as defense-in-depth for RL sandboxes. He raises follow-up design questions: could you checkpoint every tool call and syscall, and is that state best represented linearly or as a graph? Forking the sandbox at any point would enable incremental re-runs and retries while keeping a fully immutable trajectory.

Related event: Using eBPF to Rein in Rogue AI Agents' Network Access(3 posts)→

Original post →

More from coding & agent

coding & agent channel →