Stored XSS in Jupyter Server via nbconvert HTML Output

Executive Summary

This vulnerability exists in Jupyter Server versions prior to 2.20, specifically in the nbconvert HTTP handlers. These handlers render user-authored notebook HTML without including a sandbox directive in their Content-Security-Policy. Because the default behavior of nbconvert.HTMLExporter does not sanitize HTML content, a notebook containing malicious HTML payloads in display_data output can trigger stored cross-site scripting (XSS). This XSS allows attackers to access cookies, gain full authority over the /api/* endpoints, and execute remote code on the kernel.

Useful 0 Not Useful 0

Detection Guidance

This vulnerability affects Jupyter Server versions up to and including 2.19.0. Detection involves identifying if your system is running a vulnerable version of jupyter_server and if the nbconvert HTTP handlers are exposed without the sandbox Content-Security-Policy (CSP) directive.

You can check the installed version of jupyter_server with the following command:

• pip show jupyter_server

Alternatively, to check the version via Python:

• python -c "import jupyter_server; print(jupyter_server.__version__)"

To detect if the nbconvert endpoints are accessible and potentially vulnerable, you can attempt to access the nbconvert HTTP handlers and inspect the Content-Security-Policy header for the presence or absence of the sandbox directive. For example, using curl:

• curl -I http://<jupyter-server-host>:<port>/nbconvert/path/to/notebook

Check the response headers for the Content-Security-Policy. If the sandbox directive is missing, the server is vulnerable.

Note that exploitation requires an authenticated user to visit a malicious notebook, so monitoring for unusual notebook HTML payloads or unexpected nbconvert outputs may also help in detection.

Useful 0 Not Useful 0

Impact Analysis

The impact of this vulnerability is severe. An attacker can exploit the stored XSS to steal authentication cookies, which can lead to unauthorized access. They can also gain full control over the Jupyter Server API endpoints and execute arbitrary code on the kernel, potentially compromising the entire system running the Jupyter Server.

Useful 0 Not Useful 0

Compliance Impact

This vulnerability allows stored cross-site scripting (XSS) with cookie access, full API authority, and remote code execution on the kernel. Such security flaws can lead to unauthorized access to sensitive data and system control, which may result in violations of data protection regulations like GDPR and HIPAA that require strict controls on data confidentiality and system integrity.

Specifically, the ability to access cookies and execute code remotely could expose personal or protected health information, undermining compliance with these standards.

Useful 0 Not Useful 0

Mitigation Strategies

The vulnerability is fixed in Jupyter Server version 2.20. To mitigate this vulnerability, you should upgrade your jupyter_server to version 2.20 or later.

Useful 0 Not Useful 0