The vulnerability exists in the createElement function within src/utils.js. The function's purpose is to create an HTML element. Before the patch, when provided with an innerHTML option, it would directly assign this value to the element's innerHTML property. The code included a weak attempt to mitigate XSS by removing <script> tags, but this is an incomplete solution that fails to address other common XSS vectors, such as attributes like onerror. The provided exploit const maliciousHTML = '<img src=x onerror="alert(document.cookie)">'; would bypass this weak filter. The fix, introduced in commit 988826e336035b39a8608182d7b73c0e3cd78c7b, replaces the unsafe innerHTML assignment with a call to DOMPurify.sanitize(), which is a robust library for preventing XSS. Therefore, the createElement function is the specific location of the vulnerability, as it's the point where the unsanitized, user-controlled input is processed and attached to the DOM.