Challenge does not have to be random

[rersozlu]

Readme - Remark section contains below text, which is not 100% correct for all cases.
The challenge must be a random value. Otherwise, your implementation might become vulnerable to replay attacks.
I have been using webauthn to sign transaction data (on top of ethereum) to verify signatures on blockchain side. I believe replay attacks can also be prevented via auto-incremental nonce which would apply challenge as a message to be signed.

Are there any points that I am missing or is this something we should point out?

[dagnelies]

Actually, I wrote it that way to reflect the wording of the RFC which mandates it. Interestingly, there was also a recent discussion challenging this aspect of the specs here: w3c/webauthn#1856

From my point of view, strict randomness is not required either. It is rather a best practice recommendation. The important thing is to not reuse challenges to avoid "eavesdropping and replay" attacks.

On another hand, as long as the RFC says MUST be randomly generated by Relying Parties I'd prefer to keep the current wording intact.