1

Prompt & content scanning

Every prompt, tool call, and model output is scanned before it reaches the model or triggers downstream execution. Runs in-process against deterministic pattern sets. No content leaves your environment.

2

Runtime enforcement

Intercepts operations at the moment of execution — network calls, file writes, process spawning — before they complete. No amount of prompt obfuscation bypasses a hook on the operation itself.

3

OS kernel backstop

Kernel-level enforcement on Linux ≥ 5.10. Operates independently of the application runtime. Cannot be reversed from application code.

4

eBPF DaemonSet (alpha)

Kubernetes only. Extends enforcement node-wide to every container on the host.

Watch it block real attacks →
hello@hedgerow.dev