Reproduce “Tracing MCP Security Vulnerabilities: Constructing and Analyzing an MCP Security Vulnerability Dataset” 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

Xu He, Sagar Patel, Shu Wang

The authors construct a curated dataset of software-level vulnerabilities in Model Context Protocol (MCP) implementations. They filter 119 initial CVE search results by checking protocol relevance, component attribution, and disclosure status, then annotate severity, vulnerability type, affected component, and root cause. The frozen snapshot contains 24 cases, including 21 rated Critical or High; this describes disclosed records, not prevalence across MCP systems. Command injection, path traversal, and remote code execution account for 16 cases. It is an initial, disclosure-driven resource, not a census or benchmark.

Summary by alphaXiv.

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