I’ve been working on a recovery mechanism for ERC-4337/7579 smart accounts that takes a different approach from the guardian model. Instead of giving recovery power to trusted third parties, it makes recovery a permissionless economic game that anyone can play, and that attackers are structurally positioned to lose.
The full writeup is here, but the short version:
A recovery is initiated by staking LockValue ETH and waiting through a LockTime challenge window. During that window, the owner can reject the attempt and confiscate the stake. If no one objects, the recovery finalizes. That’s the core loop. It’s not a new pattern, but applying it to account custody changes the trust model in a way I think is worth discussing.
The pure mechanism alone is not sufficient, so the writeup adds several layers on top: anti-front-running commitments, and hidden watchtowers with anonymous vetoes to deal with informed attackers who target inactive owners.
This is still at the design stage, there’s no implementation yet. I wanted to collect feedback on the idea itself before starting to build.
So, what do you think of this system? Could it coexist with other recovery mechanisms? If it were available as described, would you activate it on your account? Any improvement proposals, or does this duplicate something that has already been tried?