CVE-2026-66882
LowCVSS 2.1Exploitation Probability (EPSS)
Low risk35th percentile - higher than 35% of all known CVEs
Summary
A reflected XSS vulnerability in the AshAuthentication library allows an attacker to inject malicious script into HTML pages generated during action confirmation or magic link sign-in. The issue occurs when a strategy has require_interaction enabled and request parameters are embedded into forms without proper escaping.
Risk Assessment
An attacker can send a crafted link to a victim, which upon opening executes a script in the context of the application using AshAuthentication. The script can steal sessions, cookies, or same-origin data, posing a serious threat to data confidentiality and integrity.
Recommendation
Immediately upgrade ash_authentication to version 4.14.2 or later (for 4.x branch) and to 5.0.0-rc.13 or later (for 5.x branch). If upgrading is not possible, consider temporarily disabling interaction-requiring features or applying additional input filtering.
Other vulnerabilities in AshAuthentication
Original NVD description (English source)
Improper Neutralization of Input During Web Page Generation (XSS) vulnerability in team-alembic AshAuthentication allows reflected cross-site scripting via the confirmation and magic link interaction forms. When a strategy is configured with require_interaction? set to true, AshAuthentication serves an intermediate HTML page asking the user to confirm the action by submitting a form. Both such pages embed a request parameter directly into a hidden input's value attribute without HTML escaping: lib/ash_authentication/add_ons/confirmation/confirmation_form.html.eex interpolates the confirm parameter, and lib/ash_authentication/strategies/magic_link/sign_in_form.html.eex interpolates the magic link token parameter. These templates are compiled with EEx.function_from_file/3 using plain <%= %> expressions, which perform no escaping, so the parameter is reflected verbatim. Neither accept handler validates the value before rendering it. AshAuthentication.AddOn.Confirmation.Plug.accept/2 only checks that a confirm key is present, and AshAuthentication.Strategy.MagicLink.Plug.accept/2 reads the parameter directly, so no token signature is verified at this stage and arbitrary attacker-supplied text reaches the template. An unauthenticated attacker can therefore craft a URL whose parameter terminates the attribute and injects markup, for example a quote followed by a <script> element. Because the accept phase is served over GET, loading the crafted link is sufficient; no form submission or prior authentication is required. The injected script executes in the origin of the application embedding AshAuthentication, giving it access to that origin's cookies, session, and same-origin responses. Since these pages are part of the authentication flow, a victim following what appears to be a legitimate confirmation or sign-in link is a plausible target. This issue affects ash_authentication: from 4.8.0 before 4.14.2 and from 5.0.0-rc.0 before 5.0.0-rc.13.