GIVE YOUR AGENT THE EVIDENCE

Set up in your agent.

One prompt to get started. Paste it into your coding agent to connect cve.new through MCP and bring vulnerability context into your workflow.

Your agent needs remote MCP support and access to https://cve.new/mcp.

Read the setup guide

GIVE YOUR AGENT THE EVIDENCE

Investigate in your agent.

Turn this advisory into a focused investigation. Ask your agent to check your project’s versions, explain the evidence, and suggest a compatible fix.

Your agent needs remote MCP support and access to https://cve.new/mcp.

Read the setup guide

GHSA-w2w5-w2pw-r929high8.7

NukeViet: Improper Neutralization of Input During Web Page Generation ('Cross-site Scripting')

Check your stack

The advisory, in context

Summary

A stored cross-site scripting (XSS) vulnerability exists in NukeViet CMS versions 4.x through 4.5.08. A low-privileged authenticated user can store a JavaScript payload in their profile's display name fields. The payload executes in the browser of any visitor — including administrators — who clicks the Reply ("Answer") link on a comment posted by that user.

Affected Component

The {COMMENT.post_name} template variable is interpolated without JavaScript-context escaping into an inline onclick handler in both comment block positions:

  • themes/default/modules/comment/comment.tpl line 27 (top-level comments)
  • themes/default/modules/comment/comment.tpl line 64 (nested/reply comments)
onclick="nv_commment_feedback(event, {COMMENT.cid}, '{COMMENT.post_name}')"

Root Cause

The first_name and last_name profile fields are sanitized with HTML numeric character references (' → ', ( → (, ) → ), / → /) via Request::_get_title() with $specialchars = true. This encoding is correct for plain HTML attribute and element contexts, but insufficient for a JavaScript string literal embedded inside an HTML attribute.

Browsers decode HTML entities in attribute values before the JavaScript engine parses the string. As a result, ' is decoded back to ', which terminates the JS string early and allows the remainder of the value to be executed as JavaScript.

The combined display name (nv_show_name_user(first_name, last_name)) is what reaches the template, giving an attacker up to ~200 encoded characters across both fields — sufficient for any practical payload.

Proof of Concept

Set first_name to the following value in profile settings (/index.php?nv=users&op=editinfo), then post any comment:

a');alert(document.domain);//

The value is stored as a');alert(document.domain);//.

When a visitor clicks the Reply link on the comment, the browser renders:

nv_commment_feedback(event, 1, 'a');alert(document.domain);// Tester')

causing alert(document.domain) to execute in the visitor's browser context.

A data-exfiltration variant (split across both name fields) navigates the victim's browser to an attacker-controlled URL carrying document.cookie as a query parameter. End-to-end verification was performed using a local listener.

Exploitation Conditions (default configuration)

Condition Default value Effect
captcha_area_comm 1 No CAPTCHA for logged-in users — payload delivery requires no CAPTCHA solve
auto_postcomm enabled Comments are published immediately without moderation
active_editinfo_censor 0 Profile edits take effect immediately without admin review
CSP script-src 'unsafe-inline' Inline onclick handlers execute normally

Any registered member can set the payload and post a comment with no additional steps.

If captcha_area_comm is set to 0, the name field of anonymous comments (modules/comment/funcs/post.php) is processed by the same get_title(..., 1) call, making exploitation possible without authentication.

Impact

An attacker with a regular user account can execute arbitrary JavaScript in the browser of any visitor who interacts with the Reply button on their comment, including site administrators.

Practical consequences include:

  • Privilege escalation via admin session hijacking — forging administrative actions (content modification, account manipulation) in the context of an authenticated admin.
  • Credential phishing — injecting a fake login form into the page.
  • Data exfiltration — reading page content and non-HttpOnly cookies.

Note: NukeViet session cookies carry the HttpOnly flag, so they are not directly readable via document.cookie; however, the above attack vectors remain fully viable.

Remediation

Preferred fix: Remove post_name from the inline handler entirely. Pass only cid to nv_commment_feedback and have the function retrieve the display name from the already-rendered DOM (e.g., the adjacent <strong class="cm_item"> element).

Alternative fix: If the value must be passed inline, encode it with json_encode($post_name) (PHP) so that the output is a properly escaped JavaScript string literal, not an HTML-entity-encoded one. HTML numeric character references must not be relied upon for JavaScript string escaping.

As a general note, the result of get_title(..., $specialchars=true) is safe for HTML element content and quoted HTML attribute values, but unsafe when placed inside a JavaScript string literal within an attribute. Other locations in the codebase using the same pattern should be audited.

Resources

  • OWASP: Cross Site Scripting Prevention — Rule 2: Attribute Encoding is Not Sufficient for JS Contexts
  • CWE-79: Improper Neutralization of Input During Web Page Generation
  • CWE-116: Improper Encoding or Escaping of Output

Affected packages

Start with the recorded fix for your release branch. Check your installed version and test compatibility before upgrading. Fixing this advisory does not rule out other vulnerabilities in the release.

Ecosystem versions

Earlier releases before 4.6.00

Fixed in 4.6.00

2 explicitly affected versions

4.0.24, 4.4.01

View affected-version evidence
{
  "ranges": [
    {
      "events": [
        {
          "introduced": "0"
        },
        {
          "fixed": "4.6.00"
        }
      ],
      "type": "ECOSYSTEM"
    }
  ],
  "versions": [
    "4.0.24",
    "4.4.01"
  ]
}

References