Executive brief
DOMPurify, a widely used library for cleaning HTML to prevent security attacks, contains a flaw when processing templates. When configured to protect against template-based attacks while returning a live document object, it fails to properly clean content inside HTML <template> tags. This could allow an attacker to bypass security filters and execute malicious scripts (Cross-Site Scripting) in applications that use template engines like Angular or Lit to render the sanitized content.
Technical details
A vulnerability exists in DOMPurify's SAFE_FOR_TEMPLATES feature when used with RETURN_DOM, RETURN_DOM_FRAGMENT, or IN_PLACE output modes. The root cause is that the _scrubTemplateExpressions function utilizes a NodeIterator and node.normalize() which do not descend into the DocumentFragment of <template>.content. An attacker can split a template expression (e.g., ${...}) across multiple adjacent text nodes using disallowed tags that are removed but whose content is kept. Because the final normalization pass ignores the template content, these split nodes are never merged and scrubbed, allowing the full expression to survive sanitization. If a downstream framework subsequently normalizes and evaluates this content, it results in arbitrary code execution. This issue is fixed in version 3.4.8.
Affected products
- cure53 DOMPurify >=3.0.0, <3.4.8
Timeline
- 2026-05-29: advisory: GitHub Security Advisory GHSA-gvmj-g25r-r7wr published
- 2026-07-23: disclosed: CVE-2026-65900 published to NVD