Critical Langflow Flaws Could Let Unauthenticated Attackers Run Code on Servers

Critical vulnerabilities in Langflow, an open-source tool for building AI agents, allow unauthenticated remote attackers to bypass authentication and execute arbitrary code. The flaws affect all releases from version 1.0.0 through 1.12.8, including an issue involving Model Context Protocol (MCP) tool enumeration and an insecure default configuration where automatic login (LANGFLOW_AUTO_LOGIN) is enabled by default. Attackers can leverage these weaknesses to gain superuser access or run code without credentials.

Source: The Hacker News

How Mangudai Discovers This Issue

Mangudai's port module discovers open services running Langflow instances on the external perimeter and analyzes service banners or HTTP response signatures to identify vulnerable application versions (1.0.0 through 1.12.8) exposed to the internet.

Continuous Autonomous AuditsMangudai scans target perimeters multiple times daily to catch configuration drift instantly.

Zero-Agent ArchitectureIdentifies exposure angles without requiring local installations or server access.

Check what your own domain exposes

A free scan lists the services, software versions, certificates and headers your domain shows to the internet, so you can see whether a flaw like this could apply to you. It reads what is visible from the outside; it does not confirm that a specific vulnerability can be exploited.

Launch Scan For My Domain