ruby-saml has a SAML authentication bypass due to namespace handling (parser differential)

The vulnerability stems from an incomplete fix for CVE-2025-25292, where ReXML and Nokogiri parse XML differently, producing entirely distinct document structures from the same input [1][2]. This parser differential means that a maliciously crafted SAML response can be interpreted as valid by one parser while the signature-checking logic uses the other, enabling a Signature Wrapping attack [2][3].

An attacker exploiting this issue does not need prior authentication; the attack is launched by sending a specially crafted SAML assertion to a ruby-saml-based service provider [2]. The attack requires network access to deliver the malicious SAML response, but no special privileges are needed beyond the ability to communicate with the target application [4].

The impact of a successful exploit is authentication bypass: an attacker can forge arbitrary SAML assertions and gain unauthorized access to any account or resource protected by the SAML authentication flow [2][3]. This bypass critically undermines the identity federation trust model that ruby-saml is designed to enforce.

The vulnerability is fixed in ruby-saml version 1.18.0, which incorporates proper XML namespace handling and validation ordering to eliminate the parser differential [1][3]. All users should upgrade immediately; versions 1.12.4 and earlier remain exploitable [1][2].