I disagree. We have seen this story before and it always ends the same way.

Counterpoint: the protocol-level solution is more elegant and avoids trust assumptions.

Saved. I will reference this thread next time someone asks me the same question.

Salud comunidad! En mi experiencia esto funciona muy bien.

I have been in the space since 2017 and honestly this feels different.

Doesn't match what I saw on testnet.

Could someone explain this to me like I am five? I am new and trying to learn.

Merci pour le partage.

Just finished a small audit of a 4337 paymaster. Three findings: signature replay on the alternate-mempool path, gas griefing through revert-bombing on simulateValidation, and a missing nonce check after self-call. Will publish the full report once the team patches.

Just did some napkin math and the numbers do not add up.