Reproduce “JavaSith: A Client-Side Framework for Analyzing Potentially Malicious Extensions in Browsers, VS Code, and NPM Packages” 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

Avihay Cohen

JavaSith introduces a client-side framework for identifying malicious code in browser extensions, VS Code extensions, and NPM packages, integrating static analysis, dynamic sandboxing with time simulation, and on-device Large Language Models. The framework successfully detected all tested malicious samples, including multi-stage and obfuscated attacks, by intercepting and logging harmful actions in an emulated environment.

Key insights

  • Client-side analysis, especially with local LLMs, enhances privacy and empowers users to independently verify software component trustworthiness.
  • Accelerating time-dependent malware triggers through a 'time machine' module is effective for detecting logic bombs and delayed payloads.
  • A multi-faceted approach, combining static, dynamic, and AI-driven analysis, offers a robust defense against sophisticated, multi-stage supply chain attacks.

Summary by alphaXiv.

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