MapLibre GL JS contains a flaw in its DOM sanitizer routine that allows an attacker to supply malicious attributes such as onload or ontoggle directly in map attribution strings. When the library processes attributes as a live NamedNodeMap while simultaneously removing them, a consecutive dangerous attribute can slip through the loop and persist in the cleanup. If a vulnerable map is rendered, the surviving attribute is inserted into innerHTML and its code executes in the visitor’s browser, enabling arbitrary script execution.

The vulnerability affects the open‑source MapLibre GL JS library before version 6.4.1. Systems that load a MapLibre GL JS build older than 6.4.1 and that accept untrusted third‑party style attribution strings or externally supplied custom attributions are at risk. Delivering map pages to browsers that render the affected map content provides the execution vector.

With a CVSS score of 10 the flaw is considered critical. The EPSS score is not available, but the risk is high because the bug is exploitable via ordinary web content. The CISA KEV listing is empty, indicating no known widespread exploitation yet, yet the flaw remains open for attackers who can insert malicious attribution data. The attack vector is the web page that renders the map; the attacker needs only to craft a malicious attribution entry, which is then processed by the affected library and executed in the victim’s browser.