Reproduce “Authorization Propagation in Agent2Agent Systems: An Experimental Security Evaluation of Multi-Hop Delegation” 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
Gabriele Donno, Kashif Mehmood, Katina Kralevska
The authors evaluate layered security checks for agent-to-agent reimbursement requests, separating authentication of the immediate caller from evidence of inherited authorization and cross-checking them. The prototype uses signed per-hop requests and shared-key authorization evidence. In a fixed four-party synthetic workflow, it rejects 14/15 adversarial conditions, but accepts an isolated delegation-path test that widens scope; it does not test semantic narrowing.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.