1초 미만 기동의 microVM 기반 AI 에이전트 격리 샌드박스 구현
Run Untrusted AI Agent Code Safely with Azure Container Apps Sandboxes
Run Untrusted AI Agent Code Safely with Azure Container Apps Sandboxes
Architecting isolated workspaces with Kasm: hardened, egress-controlled, disposable
권한 프롬프트 84% 감소시킨 환경 계층 기반 Agent Containment 설계
Pipelock Agent Egress Control: the missing CI primitive for AI agents
Politeness vs Enforcement: Why "Set HTTPS_PROXY" Isn't a Security Control
The Three-UID Containment Pattern for AI Agents on Linux