Similar Items: Constraining Host-Level Abuse in Self-Hosted Computer-Use Agents via TEE-Backed Isolation
- KingsGuard: Enclave Data Protection Under Real-World TEE Vulnerabilities
- Semia: Auditing Agent Skills via Constraint-Guided Representation Synthesis
- Self-Adaptive Multi-Agent LLM-Based Security Pattern Selection for IoT Systems
- Combating Organized Platform Abuse: Amplifying Weak Risk Signals with Structural Information
- MAGE: Safeguarding LLM Agents against Long-Horizon Threats via Shadow Memory
- PragLocker: Protecting Agent Intellectual Property in Untrusted Deployments via Non-Portable Prompts