Can AI Remediate Backend Failures Safely? GuardedAct with Blast-Radius-Aware Sandboxing

Wanrong Cai, Tianyu Yu, Shaorui Pi +2

Large Language Models (LLMs) have shown promising capabilities in generating remediation actions for microservice failures. However, directly executing AI-generated repair actions…