Command Injection Vulnerability in Markdownify Model Context Protocol Server

CVE-2025-58358

What is CVE-2025-58358?

Markdownify, a Model Context Protocol server designed for converting various inputs to Markdown, is susceptible to a command injection vulnerability in versions prior to 0.0.2. This issue arises from inadequate sanitization of input parameters during the execution of child_process.exec, allowing attackers to execute arbitrary system commands. If leveraged successfully, this vulnerability can facilitate remote code execution under the privileges of the server process. The server's construction and execution of shell commands with unvalidated user input significantly increase the risk of shell metacharacter injection (such as |, >, &&, etc.). The issue has been mitigated in version 0.0.2.

References