Welcome to the Jungle: What We Found Inside 15,465 Public MCP Servers

Researchers at OX Security discovered critical vulnerabilities in Anthropic's Model Context Protocol, a standard for connecting AI models and tools. The protocol, which aimed to become a universal standard, was widely adopted with thousands of developers building servers and enterprises integrating them into workflows. However, the ecosystem surrounding the protocol has significant security shortcomings. The vulnerabilities were found in Anthropic's implementation of the protocol. The discovery was made earlier this year by the OX Security team. The protocol's vulnerabilities pose a risk to the thousands of servers that have been built using it.