8 papers
Verify Smarter, Evolve Further: Efficient Harness Evolution through Behavior-Aware Verification
Jinghan Xu, Yikai Zhang, Aili Chen +3
Agent harnesses shape how language-model agents use instructions, tools, and runtime components, but adapting these harnesses requires costly verification. Existing propose-and-ver…
Convergent Detour Hijacking: Task-Preserving Resource Amplification in Skill-Based LLM Agents
Junliang Liu, Ruoyu Li, Wenxin Tang +4
LLM agents increasingly rely on third-party skills, using natural-language descriptions for selection and instruction bodies for planning. This progressive-disclosure design expose…
SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring
Yuling Shi, Jinghan Xu, Kelin Fu +12
As AI coding agents take on increasingly complex, long-horizon software engineering tasks, existing benchmarks are rapidly saturating and their evaluation quality has come under se…
Selection-Invariant Communication Compilers for Privacy-Aware Multi-Agent LLM Workflows
Jinghan Xu, Longze Fan, Zeyuan Wang +2
Structured multi-agent workflows exchange intermediate messages whose content and form can reveal private state even when the final output is safe. We identify selection-channel le…
MNC: Scope-Bound Semantic Declassification for Private LLM-Agent Communication
Jinghan Xu, Longze Fan, Zeyuan Wang +2
Multi-agent large language model (LLM) systems can expose protected state through internal messages, tool arguments, logs, and persistent memory even when their public outputs appe…
Beyond Single-Use Tokens: Durable Authorization State for Replay-Resistant LLM Agent Actions
Jinghan Xu, Longze Fan, Zeyuan Wang +2
Tool-using large language model agents frequently replan, retry failed operations, delegate tasks, and resume after crashes. These behaviors can cause one user authorization to be…
Policy-Backed Selective Regeneration under Tainted Inter-Agent Communication
Jinghan Xu, Longze Fan, Zeyuan Wang +2
Inter-agent communication is essential to multi-agent language-model systems, yet a single message may combine task-critical information with instructions not authorized by the ori…
Memory Provenance Laundering in LLM Agents: A Non-Amplification Firewall for Persistent Memory
Jinghan Xu, Yiyong Xiao, Wanru Shao +2
Long-term memory lets large language model(LLM) agents reuse prior preferences and work flows, but it also turns untrusted observations into persistent action context. We identify…