Executive brief
IBM Langflow Desktop, a tool used for building and testing AI workflows, contains a security flaw that allows users to inject malicious scripts into the application's interface. An attacker with basic access could use this to steal login credentials or hijack the sessions of other users. This could lead to unauthorized access to sensitive AI models or corporate data managed within the platform.
Technical details
A stored cross-site scripting (XSS) vulnerability exists in IBM Langflow Desktop versions 1.6.0 through 1.8.4. The flaw is located in the Markdown rendering pipeline, specifically due to the unsafe use of 'rehypeRaw' without proper sanitization during event-streaming in the Playground component. An authenticated attacker can inject arbitrary HTML or JavaScript code that executes in the context of other users' browsers. This can be leveraged to disclose sensitive credentials or perform account takeover. While the advisory mentions CWE-89 (SQL Injection), the descriptive text and title explicitly confirm this is a stored XSS vulnerability. Users should upgrade to version 1.9.0 or newer to resolve the issue.
Affected products
- IBM Langflow Desktop 1.6.0 - 1.8.4
Timeline
- 2026-04-28: disclosed: Initial publication by IBM
- 2026-04-30: advisory: NVD publication date