Reproduce “Treating the Model as an Untrusted Component: Deterministic Policy Enforcement, Human Approval and Taint Tracking Against Prompt Injection in Tool-Using Agents, with Validation on Open-Weights Models” 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

Mustafa S. Aljumaily, Nawar S. Alseelawi, Raghad Alshabandar

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.