Executive brief
Mage AI, an open-source data pipeline tool, contains a security flaw in its sign-in process. An attacker can create a malicious link that, if clicked by a user, executes unauthorized code in the user's browser. This could allow an attacker to steal login sessions, access sensitive data, or perform actions on behalf of the victim within the Mage AI platform.
Technical details
A DOM-based Cross-Site Scripting (XSS) vulnerability exists in the Mage AI SignForm component (mage_ai/frontend/components/Sessions/SignForm/index.tsx). The application extracts the 'redirect_url' parameter from the URL and directly assigns it to 'window.location.href' within the 'onSuccess' callback of the 'useMutation' hook without validation or sanitization. An attacker can exploit this by crafting a URL using the 'javascript:' protocol. For authenticated users, the script executes immediately upon visiting the link; for unauthenticated users, the script executes immediately after they complete the login or registration process. This can lead to session hijacking and full account takeover. As of the advisory date, the vendor has not responded to disclosure attempts.
Affected products
- Mage AI Mage AI up to 0.9.79
Timeline
- 2026-03-07: disclosed: Vulnerability reported to vendor
- 2026-06-06: advisory: Public disclosure via VulDB and NVD