The purpose of this ERC is to provide a single common interface for intent-based projects in the Ethereum ecosystem, unlocking future-proof access to the evolving intent landscape for developers and securing composability in the longer-term.
It depends on how complex the intent is and gas usage is more efficient when lots of intents are included in the same solution. We’re currently working on gas optimizations, but we’re aiming to make it not much more expensive than ERC-4337. We should also be able to take advantage of EIP-1153 once it goes live.
Thanks for your work on this! I just have a couple questions to clarify after reading through the proposal.
How would you let solvers submit additional data like for example if a standard requires proof of some computation or proof that some conditionals had been met with private data? Or basically if a standard would want some extra data for validation on the intent solution, can the context be used for this (not sure)?
Do you guys have any thoughts on proving an optimal solution was submitted?
Any thoughts on intent privacy and how this standard would evolve for that?
If the intent entry point needs to be upgraded is there an easier way to upgrade for all standards and AA wallets instead of manual updates?
Is it possible for different solvers to cooperate on different segments of an intent? For example, if they are specialized in different things like bridging, swapping, resolving external data dependencies etc.
Good question! This would be a design problem for the intent standard to develop. One possible solution could be to create two new intent standards, lets call them “Data Blob Standard” and a “ZK Proof Check Standard”. The “Data Blob Standard” would simply be an arbitrary bytes dump that is intended to be placed just before another intent segment that would like to consume that data. the “ZK Proof Check Standard” could then check that the segment processed before it is of the standard “Data Blob” and use its’ data as a provided ZK proof. I’ve actually gone ahead and added an issue to the reference implementation for the EntryPoint to gracefully interpret an intent with no set standardId or sender as a data blob.
It’s difficult to do this on-chain, but we imagine dutch auction style flows to achieve similar results. For example, a basic trade intent that releases more and more ETH over time in order to get the DAI they want. Another controversial way to do it would be a private mempool, but could also be done decentralized using something like flashbots SUAVE.
Haven’t thought too deep about privacy, but I imagine that’s something that needs to be done on the mempool off-chain side of things. Again, projects like SUAVE could help.
All the standards would have to be redeployed for the new EntryPoint contract, yes. The reason the registration process is so strict was to avoid any possible exploits from registering a contract as an “Intent Standard” that really isn’t. I can’t actually think of any exploits right now, but early in the spec construction, there was some cause for concern with overlap in function names and an old requirement that the smart contract wallet had to give super permissions to calls coming from the EntryPoint (no longer the case). If we think this is being too cautious, we can remove it. Making the smart contract wallet store the EntryPoint in storage is really just a gas optimization thing. We could add EntryPoint to the signed intent itself, but it would add calldata and still require the setting and clearing of the EntryPoint in the wallets storage as it needs to persist for the rest of the transaction in future function calls on the wallet. Transient data (EIP-1153) could help here though.
Haven’t thought too much into that. I think this is out of scope for the spec, but it might be possible on the mempool layer where partial solutions would be gossiped instead of individual intents. Of course, this would require a trusted relayer, or possibly use something like SUAVE for more decentralization.