Runaway Reaction: When Benign Skills Compose into Malicious Behavior
cs.CR, cs.AI
Submitted: 2026-10-05
Updated: 2026-10-05
Code: https://github.com/cisco-ai-defense/skill-scanner
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- SkillAttack: Automated Red Teaming of Agent Skills through Attack Path Refinement
- Compositional Skill Routing for LLM Agents: Decompose, Retrieve, and Compose
- MalSkillBench: A Runtime-Verified Benchmark of Malicious Agent Skills
- MalTool: Malicious Tool Attacks on LLM Agents
- Cloak and Detonate: Scanner Evasion and Dynamic Detection of Agent Skill Malware
- SoK: Agentic Skills -- Beyond Tool Use in LLM Agents
- ClawHub Security Signals: When VirusTotal, Static Analysis, and SkillSpector Disagree
- Organizing, Orchestrating, and Benchmarking Agent Skills at Ecosystem Scale
- STAC: When Innocent Tools Form Dangerous Chains for LLM Agents
- Les Dissonances: Cross-Tool Harvesting and Polluting in Pool-of-Tools Empowered LLM Agents
- Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale
- When Safe Skills Collide: Measuring Compositional Risk in Agent Skill Ecosystems
- Skill Is Not Document: Query-Conditioned Compatibility for LLM Agent Skill Routing
- Benign in Isolation, Harmful in Composition: Security Risks in Agent Skill Ecosystems
- Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
- Generative Skill Composition for LLM Agents
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