I can’t help but feel that this project is missing the point of a cryptography library entirely.

Crypto libraries must foremost be correct and safe. Performance is nice but secondary. I’m sure there is a lot of potential to speed up libsecp by using assembly etc., but vibecoding hundreds of commits in a few days without review completely undermines confidence in correctness and safety. I would expect people to generally react by keeping a project like this at pike’s length.

Also, licensing it under AGPL is likely going to make it a non-starter for most potential users.