Yes, you are correct; U2F hardware keys can’t be used to log into sites that don’t support them. (That’s why I prefaced the remark you quoted with the word “Ideally,” which you did not include in your quote.)
Nevertheless, U2F may still help prevent unauthorized logins to your account on sites that do not support U2F. (Please note that I said “may”—as in “under some circumstances”—and not “will.”)
Did you read either of the articles that I linked to? Here’s the TL;DR synopsis.
Even if your Enpass password can’t be cracked, the passwords (to websites that do not support U2F) in your vault are still vulnerable to phishing attacks on YOU, the person who knows the Enpass password.
If such an attack involves a MITM, then U2F may thwart the attack and thus help secure your password to a website that does not support U2F.
For details, please see this article that I cited and scroll down to “U2F and Security Keys,” which begins with:
If the human is the biggest vulnerability in a phishing attack, then we should just remove the human in the process.
This is what U2F tries to do. We relieve the human the burden of identifying between fake and real sites. This is going to be taken care of by the YubiKey and the browser working together.
And later on:
The browser checks the certificates of the website, before it asks the security key to generate any codes. It makes sure doesn’t allow our fake websitelastpass.com.es
to get codes for lastpass.com
. This makes it difficult for a hacker to do a MITM attack. The site not only has to look like the legitimate site, but it also has to have to correct certificates.
And that’s why, in the article I cited, Twillio, which used OTP, was breached in August 2022 but Cloudflare, which used U2F, wasn’t.
If the password to the vault is good and the cryptography is implemented correctly, then the human who knows the vault password is the weak link. All that’s needed is a website login page that looks legitimate and this human will reveal the sought-after website password (even without a wrench
).
In such cases, the risk is not whether the vault is in the cloud; the vault can not be cracked (without a quantum computer),
The risk is the human being phished.