Open madwizard-thomas opened 6 years ago
I've been thinking about that. I don't think it's direct attack, because the result messages contain a random challenge already (as part of the FIDO / WebAuthn protocol). But you may be right that the day a risk engine gets hooked up to this, it's possible to inject noise / increase risk scores through CSRF. Not an attack on its own, but perhaps part of a broader broader attack.
I aspire to write a XACML-style policy engine and an risk engine some day, so maybe this comes in to play as part of that.
Would it be necessary to protect the POST requests in this library with a CSRF token? The response can't be read from other security contexts but it might still be abused to wrongly trigger risk engines.