G-Safeguard: A Topology-Guided Security Lens and Treatment on LLM-based Multi-agent Systems

Open in new window