Send a test copy of one step's email

Sends a copy of one send_email step's email to addresses you name, so somebody can look at it. Needs sending: write, and is a write.

It touches nothing else: no run, no delivery record, no status change, nobody on the list. Run it twice and the only thing that happens twice is somebody reading the email.

It matters more than an article's test send. An article is written, looked at and sent within an afternoon. An automation email is written once and then sent, unattended, for months, to people who arrive one at a time, so there is no moment at which anybody watches it go out. Without this the first person to see it rendered is a subscriber.

It renders through the same renderer and the same personalization pass the real send uses, so what lands is what a subscriber would get, with the placeholders filled in from example values and the unsubscribe link pointing at a page that explains itself rather than at a live token.

Three gates, not five. There has to be a step, an email on it, and a verified address to send from. The footer scan and the image probe do not apply: looking at the email is how you notice the footer has gone, and both already ran when the graph was saved and run again when the automation is turned on.

Left out, to sends the copy to the address on the account this credential belongs to. That address is not echoed in the answer, which says sent_to_owner instead: it is revealed only on GET /me, under account: read.

Charged to the same daily cap an article's test send is, counted per credential, and reported as a 429 that is not one of the RateLimit-Policy budgets. These two are the only operations here that mail somebody who never subscribed.

Publishes no event.