Reproduce “Mutate to Bypass: Autonomous Endpoint Evasion via Knowledge-Driven Multi-Agent Orchestration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Weifeng Yuan, Wenbo Guo, Qingyun Du, Jun Chen, Feng Dong, Haoyu Wang, Yang Liu
AutoBypass turns security knowledge into shellcode-loader samples using a knowledge base, specialist agents, and iterative endpoint testing. It uses observed failures to guide refinements rather than discover new evasion techniques. Across seven products, GPT-5.1 achieved 90.0% success against Windows Defender with Cobalt Strike shellcode. For Llama-3.1-8B, full knowledge base success was 80.0%, versus 33.3% without it. Results concern tested shellcode-loader tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.