Version 1.12.0
Right-click the code field on any site and pick one item. The code goes in. No popup, no switching windows, no reading six digits off a screen and typing them before they expire.
The menu item appears on text fields. Nothing else changes.
Inserting a code into a page is the one thing this extension does that touches a page at all. It is worth being precise about what that means, because “it can type into websites” and “it can read websites” are very different powers, and only the first one is being asked for.
This is not a content script. No code from the extension exists on any page you visit. It is injected only when you choose the menu item, and what that gesture grants is a single tab, a single invocation, revoked as soon as you navigate. Nothing is left behind afterwards.
The extension declares no host permissions at all — you can check the list in the Chrome Web Store, it is four short entries and none of them is a website. It cannot read the pages you have open, and it does not know which sites you visit.
It looks at the input fields on the page for one purpose: to decide which one should receive the code. What it finds is used there and then and goes nowhere else. The extension makes no network request while doing this — there is no server to send anything to.
If there is no field it can be confident about, it copies the code to your clipboard and says so, instead of putting it somewhere wrong. If a field is too short to hold the whole code, it declines — a truncated code looks like success and fails at the worst moment.
None of this has to be taken on trust. The entire page-side implementation is one readable file, and it is the only code that ever runs in a page: read it on GitHub.
The full technical changelog is on GitHub: release notes.