Intents
How a user says what should happen without touching the funds.
An intent is you saying what should happen. Pay this person. Swap this for that. Call this contract. It's your entire involvement in a payment, and on its own it has no power at all.
Saying, not signing money away
With a wallet, approving a payment and moving the money are the same action. Your signature is what makes the transfer valid.
Here they're separate. The intent says what you want. The spending right says you're allowed. Neither does anything alone.
That split is what lets you never touch the money. You're describing an outcome, not signing tokens out of something you hold, because you don't hold anything.
Saying it and being allowed to are separate. With a wallet, your signature is both at once. Here the intent describes the outcome and the right permits it, and you need both. That split is what lets you never touch the money.
An intent by itself is harmless
An intent with no valid right behind it does nothing. It isn't half-approved or sitting in a queue. There's simply nothing to execute against. Which means intents can be passed around fairly openly: they're not bearer instruments and they carry no authority.
What an intent has to say
- The action. Pay, swap, execute.
- The details. Who, how much, which asset, and whatever else the action needs.
- A binding. Enough to tie this intent to the right being used, so a valid right can't be pointed at a different outcome than the one you approved.
That last one is the part that has to be right. If an intent weren't tied to the right being used, anyone who saw a valid proof could reuse it to pay themselves instead. The binding is what makes a proof useless for anything except the exact thing it was made for.
Not decided yet. The intent format, how you submit one, how it's bound to a right, and whether you can batch or schedule them.