14 papers
SKILLLITE: Evidence-Guided Malicious Skill Auditing with Compact LLMs
Haoran Ou, Gelei Deng, Xuanye Zhang +3
As LLM-based agents perform increasingly complex tasks, Agent Skills have emerged as a flexible mechanism for extending their capabilities. An Agent Skill packages task-specific in…
Mutate to Bypass: Autonomous Endpoint Evasion via Knowledge-Driven Multi-Agent Orchestration
Weifeng Yuan, Wenbo Guo, Qingyun Du +4
Public reports and open-source resources expose many EDR evasion techniques, but it remains unclear whether commercial Endpoint Detection and Response (EDR) systems can withstand t…
Skills That Don't Exist: A Large-Scale Study of Hallucinated Skill Recommendation in LLM Agents
Weifeng Yuan, Wenbo Guo, Feng Dong +2
LLM agents acquire new capabilities by downloading skills from open registries. Instead of browsing these catalogs manually, developers typically ask the agent to recommend and ins…
When Do Intrinsic Rewards Work for Code Reasoning? A Comprehensive Study
Xiaolong Jin, Xuandong Zhao, Wenbo Guo +2
Reinforcement learning with verifiable rewards (RLVR) has driven substantial progress in large language model reasoning, but relies on ground-truth supervision that is costly or in…
Seeing Is Not Screening: Multimodal Hidden Instruction Attacks on Agent Skill Scanners
Xiaojun Jia, Jie Liao, Simeng Qin +5
Agent skills are emerging as an important attack surface in LLM-based systems. Through an empirical study of existing skill scanners, we find that current defenses primarily rely o…
MalwarePT: A Binary-Level Foundation Model for Malware Analysis
Saastha Vasan, Yuzhou Nie, Kaie Chen +6
Automated malware analysis increasingly relies on machine learning, yet most existing methods remain task-specific and depend on handcrafted features or narrowly scoped models. Rec…
OpenSage: Self-programming Agent Generation Engine
Hongwei Li, Zhun Wang, Qinrun Dai +11
Agent development kits (ADKs) provide effective platforms and tooling for constructing agents, and their designs are critical to the constructed agents' performance, especially the…
TermiGen: High-Fidelity Environment and Robust Trajectory Synthesis for Terminal Agents
Kaijie Zhu, Yuzhou Nie, Yijiang Li +10
Executing complex terminal tasks remains a significant challenge for open-weight LLMs, constrained by two fundamental limitations. First, high-fidelity, executable training environ…
rePIRL: Learn PRM with Inverse RL for LLM Reasoning
Xian Wu, Kaijie Zhu, Ying Zhang +2
Process rewards have been widely used in deep reinforcement learning to improve training efficiency, reduce variance, and prevent reward hacking. In LLM reasoning, existing works a…
To Defend Against Cyber Attacks, We Must Teach AI Agents to Hack
Terry Yue Zhuo, Yangruibo Ding, Wenbo Guo +1
For over a decade, cybersecurity has relied on human labor scarcity to limit attackers to high-value targets manually or generic automated attacks at scale. Building sophisticated…
DevOps-Gym: Benchmarking AI Agents in Software DevOps Cycle
Yuheng Tang, Kaijie Zhu, Bonan Ruan +14
Even though demonstrating extraordinary capabilities in code generation and software issue resolving, AI agents' capabilities in the full software DevOps cycle are still unknown. D…
PromptArmor: Simple yet Effective Prompt Injection Defenses
Tianneng Shi, Kaijie Zhu, Zhun Wang +13
Despite their potential, recent research has demonstrated that LLM agents are vulnerable to prompt injection attacks, where malicious prompts are injected into the agent's input, c…
Co-PatcheR: Collaborative Software Patching with Component(s)-specific Small Reasoning Models
Yuheng Tang, Hongwei Li, Kaijie Zhu +3
Motivated by the success of general-purpose large language models (LLMs) in software patching, recent works started to train specialized patching models. Most works trained one mod…
MELON: Provable Defense Against Indirect Prompt Injection Attacks in AI Agents
Kaijie Zhu, Xianjun Yang, Jindong Wang +2
Recent research has explored that LLM agents are vulnerable to indirect prompt injection (IPI) attacks, where malicious tasks embedded in tool-retrieved information can redirect th…