JupyterLab Desktop stores user‑supplied remote‑server URLs and RSS news‑feed titles in the Welcome renderer without proper encoding, creating a cross‑site scripting flaw. An attacker who convinces a user to submit a crafted URL can inject arbitrary JavaScript that runs inside the Welcome view, and because the view has access to a privileged inter‑process communication bridge, the attacker can execute code with the same privileges as the user. The same mechanism can be triggered by controlling the content of the Jupyter Blog RSS feed, but that requires control over the feed source or cached data. The vulnerability maps to CWE-79 (Cross‑Site Scripting), CWE-116 (Improper Encoding of Output), and CWE-94 (Code Injection). If exploited, the attacker can gain complete control of the victim’s machine.

All supported platforms running JupyterLab Desktop prior to version 4.6.2‑1 are affected. The product is identified by the CNA as jupyterlab:jupyterlab‑desktop. No further platform or operating‑system distinctions are supplied. The fix has been released in version 4.6.2‑1 and later.

The CVSS score of 9.4 indicates a critical severity and the lack of an EPSS entry suggests limited public exploitation data but a high risk for affected users. The vulnerability is not catalogued in the CISA KEV list. Attackers can exploit the flaw through user interaction by convincing a user to load a malicious URL or by compromising the RSS feed source. Since the malicious payload is executed with the user’s privileges, it enables full control of the victim’s desktop environment.