Browser based email wallet connector

zkemail / email-wallet

A smart contract wallet controlled using email

https://emailwallet.org

MIT License

131 stars 16 forks source link

Browser based email wallet connector #45

Open saleel opened 9 months ago

saleel commented 9 months ago

Allow dApp websites to easily integrate with EmailWallet by installing a JS script.

The JS script will

inject window.ethereum and listen to calls made to it based on the spec.
intercept transactions and generate calldata that matches Email Wallet Execute operation.
Generate a mail to link with subject like `Execute 0x{calldata}.
Open the mailto link.

The ideal behaviour for a Uniswap UI would be

User clicks "Use Email Wallet" button
UI reflects to show connected status
User enter amounts to swap, and click "Swap" button
Their email app opens with a new email to the relayer with subject "Execute: 0xababa..."
Once they send the email, the desired Swap happens.

There is a MVP / WIP in connect-email-wallet branch.

Divide-By-0 commented 9 months ago

So what's missing in the browser-connector branch?

saleel commented 9 months ago

Its only a MVP that opens the mailto link for a target+call data. We want a script that others can include in their UI that implements the functionality in generic way - ideally confirming to https://eips.ethereum.org/EIPS/eip-1193

Divide-By-0 commented 4 months ago

This can also be done by a chrome extension that injects itself into login flows.

saleel commented 4 months ago

It would be better if users don't have to install any extensions. Dapps can install our JS SDK to interact with user's email wallet.

In "browser-connector" model we open the mailto link for each transaction the user want to make in the dapp; or we send an email to the user and ask them to reply.
In "email-oauth" model we do email send/reply only once and then use ephemeral keys.

(both are kind of oauth as in "Sign in with Email Wallet", just differs in the flow/security model)

Anyway, implementation of browser-connector is not complete in this branch. There is a rough MVP, but there is no SDK that others can integrate yet. We can probably close this issue, and open a new one later based on which of the above approach we are going to take.