文章摘要
Security researchers have uncovered three critical vulnerabilities impacting Anthropic's Model Context Protocol (MCP) Git server. These flaws reportedly include an authentication bypass, a command injection vulnerability, and an information disclosure issue. The identified vulnerabilities could potentially allow unauthorized access to sensitive MCP-related code repositories and facilitate remote code execution on the server. Anthropic has acknowledged these security concerns and subsequently released patches to address the vulnerabilities, advising all users and developers leveraging their MCP Git server to apply the updates without delay. This highlights ongoing security considerations within core AI protocol infrastructure.