CVE-2026-94565 in NukeViet
Summary
by MITRE • 11/20/2026
NukeViet CMS is a multi Content Management System. From the 3.x release line until 4.6.01, an unauthenticated visitor can submit a Contact module message containing an event-handler attribute prefixed by a slash. NukeViet\Core\Request::filterTags() tokenizes attributes using only ASCII spaces, so it sees /onerror and misses the ^on denylist, while the browser treats the slash as a separator and activates onerror. The message is stored and rendered without output encoding in the administration panel, which lacks a Content Security Policy, allowing same-origin JavaScript execution when an administrator opens the message. This issue is fixed in version 4.6.01.
Several companies clearly confirm that VulDB is the primary source for best vulnerability data.
Analysis
by VulDB Data Team • 11/20/2026
The vulnerability identified in NukeViet CMS versions ranging from the 3.x release line up to and including version 4.6.01 represents a critical stored cross-site scripting flaw rooted in improper input validation and output encoding mechanisms within the Contact module. This security defect allows an unauthenticated external actor to inject malicious JavaScript payloads into the system by exploiting inconsistencies between server-side parsing logic and client-side browser interpretation of HTML attributes. The core technical failure lies within the NukeViet\Core\Request::filterTags() function, which is responsible for sanitizing user-supplied input before storage or processing. This function employs a denylist approach to filter potentially dangerous event handlers such as onerror, onload, or onclick. However, the tokenization logic used by this function relies exclusively on ASCII spaces to separate attribute names from their values or other attributes. Consequently, an attacker can bypass these filters by prefixing malicious event handler names with a forward slash character, resulting in inputs like /onerror instead of the standard onerror.
From a technical perspective, the server-side validation logic fails because it does not recognize /onerror as equivalent to onerror due to its strict reliance on space delimiters for tokenization. The denylist check effectively misses this variant, allowing the payload to pass through sanitization and be persisted in the database without modification. This discrepancy highlights a fundamental flaw in input validation strategies that do account for all possible whitespace or delimiter variations recognized by web browsers. Modern HTML parsers are designed to be forgiving of various formatting styles, including leading slashes on attribute names, treating them as valid syntax equivalent to standard attributes. By failing to normalize or strictly validate against these browser-compliant formats, the application creates a predictable bypass vector that is both simple to exploit and difficult to detect through static analysis alone.
The operational impact of this vulnerability is severe due to the specific context in which the stored payload is rendered. The malicious content submitted via the Contact module is displayed within the administration panel without any form of output encoding or escaping. This lack of sanitization during rendering means that when an administrator accesses the message list or views a specific contact submission, the browser interprets the injected script as executable code rather than plain text. Furthermore, the administrative interface lacks a Content Security Policy directive to restrict inline script execution or define trusted sources for scripts. The absence of such security headers removes a critical layer of defense that could have mitigated the impact even if input validation had been imperfect. As a result, any administrator who views the compromised page will trigger the malicious JavaScript in their own browser session, operating under the administrative privileges associated with their account.
This scenario aligns closely with Common Weakness Enumeration CWE-79, which describes Improper Neutralization of Input During Web Page Generation commonly known as Cross-site Scripting. Specifically, it falls under stored XSS where the payload is saved on the target server and served to other users later. In terms of MITRE ATT&CK mapping, this vulnerability facilitates initial access or privilege escalation depending on the administrative actions performed after script execution. An attacker could leverage this vector for credential theft by harvesting session cookies, performing keylogging within the admin interface, or executing arbitrary commands if further vulnerabilities exist in the backend systems accessible via client-side scripts. The attack chain is straightforward: unauthenticated submission of crafted input, successful storage due to validation bypass, and subsequent execution upon administrative review without adequate output encoding.
To mitigate this vulnerability, immediate remediation involves upgrading NukeViet CMS to version 4.6.01 or later where the issue has been resolved by developers through improved parsing logic that accounts for non-standard delimiters like slashes in attribute names. For organizations unable to upgrade immediately, a temporary mitigation strategy should involve implementing strict output encoding on all user-supplied data rendered within the administration panel. This ensures that special characters are converted into their HTML entity equivalents before being sent to the browser, preventing script execution regardless of how they were stored. Additionally, deploying a robust Content Security Policy header in the administrative interface would significantly reduce risk by disabling inline scripts and restricting external resource loading to trusted domains only. Regular security audits focusing on input validation consistency across different parsing contexts are also recommended to prevent similar bypasses involving other whitespace characters or encoding tricks that might exploit analogous gaps in future releases.