Secure command sandbox with ACL-based filtering preventing AI agents executing destructive operations like file deletions during panic states.
nodejs typescript command-execution child-process process-management ai-safety ai-agents deno allowlist denylist security-validation llm-tools terminal-security path-traversal-prevention agent-protection command-sandbox destructive-command-blocking goal-frustration-collapse shell-injection-protection workspace-sandboxing
-
Updated
May 21, 2026 - TypeScript