Research library
Browse by topic.
Explore the collection by research area.
Research areas
2686 papers
Governance and Policy
- I-HITL: Inalienable Human-in-the-Loop
- Then there is prompt injection, a malicious email, ...: The Wild Wide West of OpenClaw Agentic System Security
- CoSA: The Conflict of Skills Attack in Real-World LLM Agent Ecosystems
1859 papers
Benchmarks and Evaluation
- The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
- CoSA: The Conflict of Skills Attack in Real-World LLM Agent Ecosystems
- CEPO: A General Framework for Improving Prompt Optimization in Multi-Agent Systems
1560 papers
Orchestration Risk
- Silent and Costly: Characterizing and Detecting Prompt-Caching Mis-Integration in LLM Agents
- Scope-Gap Injection: Exploiting Evaluator Channel Blind Spots in the OpenClaw Agent Security Competition
- Babysitting AI Agents: A User-Centered Taxonomy and Study of Agentic Delegation Failures
1519 papers
Trust and Identity
- I-HITL: Inalienable Human-in-the-Loop
- Then there is prompt injection, a malicious email, ...: The Wild Wide West of OpenClaw Agentic System Security
- Babysitting AI Agents: A User-Centered Taxonomy and Study of Agentic Delegation Failures
1418 papers
Agent-to-Agent Communication
- The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
- CEPO: A General Framework for Improving Prompt Optimization in Multi-Agent Systems
- AgentFlowShield: Defending LLM Agents Against Metadata Side Channels
827 papers
Prompt Injection
- The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
- Then there is prompt injection, a malicious email, ...: The Wild Wide West of OpenClaw Agentic System Security
- CoSA: The Conflict of Skills Attack in Real-World LLM Agent Ecosystems
434 papers
Memory Poisoning
- Provenance-Gated Memory and Deontic Output Supervision for Epistemically Grounded Autonomous Agents
- LLM WikiLeaks: Uncovering URL-Based Capability Extension and Security Circumvention in Agentic Message Boards
- EvolveMAS: A Unified Self-Evolution Framework for Decentralized Multi-Agent Systems