Remark42's image proxy fetches remote URLs and determines if the resource is an image by inspecting only the Content-Type header during download, but during serving, it derives the Content-Type by sniffing the actual bytes. This inconsistency allows an attacker to host a URL that claims to be an image (e.g., Content-Type: image/png) but returns HTML/JavaScript content. The proxy accepts the resource as an image during download but serves it with a text/html content-type after sniffing, causing the victim's browser to execute attacker-controlled scripts within Remark42's origin. No Remark42 account is needed for exploitation; the attacker only needs to host the malicious URL and deliver the proxy link to a victim. The vulnerability affects Remark42 versions 1.6.0 through 1.15.0 and is fixed in 1.16.0.

Successful exploitation allows an attacker to execute arbitrary JavaScript in the context of the Remark42 origin, leading to potential theft of sensitive information, session hijacking, or other actions permitted by the victim's browser security context. The vulnerability requires user interaction (clicking or visiting the malicious proxy link) but no authentication. The CVSS score of 8.2 reflects high severity with network attack vector, low attack complexity, no privileges required, user interaction required, scope change, high confidentiality impact, low integrity impact, and no availability impact.

Upgrade Remark42 to version 1.16.0 or later, where this vulnerability is fixed. No other official remediation or temporary fixes are documented. Until upgrading, avoid clicking or distributing untrusted proxy links that fetch remote images through Remark42's image proxy.