Reproduce “LLM-Based Penetration Testing in the Presence of Honeypots” 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
Xinhong Xie, Piyush Nagasubramaniam, Neeraj Karamchandani, Sencun Zhu
The authors introduce a policy for allocating an LLM penetration-testing agent’s limited inference budget across candidate hosts. It screens and ranks hosts using pre-entry honeypot evidence, then checks footholds and abandons suspicious targets. In a $5 test, it confirmed 3.66 ± 0.47 of 12 genuine hosts, but showed no advantage at $1; results are limited to a 20-host post-triage testbed.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.