Cross-site scripting in secure_headers - CVE-2026-54163
Published: November 6, 2026
Vulnerability details
The vulnerability allows an unauthenticated remote attacker to bypass Content Security Policy restrictions, enable cross-site scripting, and exfiltrate CSP violation reports.
The vulnerability exists due to secure_headers interpolating values for sandbox, plugin_types, and report_to into CSP headers without removing semicolons, carriage returns, or line feeds when applications forward untrusted input through per-controller override or append APIs, while validation accepts sandbox values beginning with allow-, broadly matching MIME types, and any nonempty reporting-group string. An unauthenticated remote attacker can inject semicolon-separated directives through these values: sandbox and plugin-type injections place a permissive script-src before the legitimate directive, overriding it under CSP's first-occurrence rule and allowing reflected or stored scripts to execute when a victim visits affected content, whereas reporting-group injection can redirect report-to or report-uri reporting to attacker infrastructure to disclose violated URLs, blocked URIs, source files, line numbers, and sample snippets, including victim-internal URLs, without enabling script execution by itself.