← Back to directory
AI SoftwareWeb linkPublished

OpenShell — Sandbox and Policy Runtime for AI Agents

The policy-enforcing shell/sandbox layer that makes autonomous AI agents like OpenClaw safe to run in production.

About this resource

OpenShell provides guardrails, YAML-based policies and sandboxed command execution (Docker, SSH or native backends) for AI terminal agents — the security runtime behind enterprise agent deployments such as NVIDIA's NemoClaw. Source: https://github.com/vercel-labs/openclaw-sandbox

OpenShellOpenClawsecuritysandbox

More in AI Software