IBM Langflow Desktop EUVD-2026-26425

An attacker must first obtain valid authentication credentials to IBM Langflow Desktop (PR:L in CVSS). The vulnerable application instance must be running a version between 1.6.0 and 1.8.4. The attacker must inject malicious JavaScript into a persistent UI element (workflow, component, or node property) that is later viewed by another authenticated user. No special configuration or non-default settings are required; the vulnerability exists in the base application's default rendering pipeline. The scope change (S:C) indicates that the impact can affect other components or users beyond the immediate vulnerable context. Exploitation does NOT require user interaction from the attacker's perspective (UI:N in CVSS), but execution of the injected payload depends on a second user viewing the malicious content, creating indirect user interaction in the overall attack chain.

CVSS 6.4 with network vector (AV:N), low complexity (AC:L), and low privilege requirement (PR:L) indicates moderate risk with relatively easy exploitation once authenticated. The scope change (S:C) extends impact beyond the vulnerable component. However, the vulnerability requires prior authentication, which limits the immediate attack surface compared to unauthenticated XSS. The stored nature (persistence across sessions) makes this more dangerous than reflected XSS. No EPSS score, POC status, or KEV listing provided in input data. The CWE-89 tag in the input appears to be a tagging error (CWE-89 is SQL Injection, not XSS - the description clearly indicates stored XSS, which is CWE-79). This discrepancy suggests either misclassification or incomplete input data; the actual vulnerability is XSS-based, not SQL injection. Real-world risk is elevated for multi-user instances where authenticated users may be social-engineered into clicking crafted links or viewing malicious workflows, but the authentication barrier and need for workflow/component interaction provide some natural friction.

An authenticated user with access to Langflow Desktop creates a workflow or component and embeds malicious JavaScript in a field (e.g., node name, description, or parameter value) that is rendered without sanitization in the Web UI. When another authenticated user views this workflow, the JavaScript executes in their browser within the application's trusted context, allowing the attacker to steal session cookies, capture credentials entered into other forms, or redirect the user to a phishing page, all appearing to come from the legitimate application.