Publications

HarnessSecurity-Bench: Do Security Mechanisms Really Protect Coding Agent Harnesses? Permalink

Published in arXiv preprint arXiv:2610.07639, 2026

An empirical study of security mechanisms in 40 coding agent harnesses, with a benchmark evaluating security and task utility through 23 tasks across five attack surfaces.

Recommended citation: Zhengyang Zhu, Liming Huang, Runmin Ji, Mingxi Ye, Zihan Zhou, Hanyang Guo, Jingwen Wu, Yuhan Ye, Yuming Feng, Hong-Ning Dai, Zibin Zheng. "HarnessSecurity-Bench: Do Security Mechanisms Really Protect Coding Agent Harnesses?" arXiv preprint arXiv:2610.07639, 2026.
PDF

Project website

VRExplorer: A Model-based Approach for Semi-Automated Testing of Virtual Reality Scenes Permalink

Published in the 40th IEEE/ACM Automated Software Engineering Conference (ASE 2025), 2025

A model-based VR testing tool built on the EAT (Entity-Action-Task) framework, achieving up to 122.8% ELOC and 52.8% method coverage improvements over the SOTA on 11 representative VR projects.

Recommended citation: Zhengyang Zhu, Hong-Ning Dai, Hanyang Guo, Zeqin Liao, Zibin Zheng. "VRExplorer: A Model-based Approach for Semi-Automated Testing of Virtual Reality Scenes." the 40th IEEE/ACM Automated Software Engineering Conference (ASE 2025).
IEEE Xplore | PDF