S cubed: Improving Agent Safety through Multi-Stage Defense
Zibo Xiao, Haoyu Wang, Jun Sun
cs.CR, cs.AI
Submitted: 2026-08-03
Code: https://github.com/FFchopon/S3Framework
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructions
- LlamaFirewall: An open source guardrail system for building secure AI agents
- A Safety and Security Framework for Real-World Agentic Systems
- Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations
- ATBench: A Diverse and Realistic Agent Trajectory Benchmark for Safety Evaluation and Diagnosis
- Towards Secure Agent Skills: Architecture, Threat Taxonomy, and Security Analysis
- SafeHarness: Lifecycle-Integrated Security Architecture for LLM-based Agent Deployment
- AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security
- Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale
- Memory Poisoning Attack and Defense on Memory Based LLM-Agents
- SafeClaw-R: Towards Safe and Secure Multi-Agent Personal Assistants
- A-MemGuard: A Proactive Defense Framework for LLM-Based Agent Memory
- AIR: Improving Agent Safety through Incident Response
- SafeAgentBench: A Benchmark for Safe Task Planning of Embodied LLM Agents
- Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM-based Agents
- AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs