edge_app_publish_from_html takes HTML and publishes it as a Screenly edge app. Ask an assistant to upload this as a Screenly app and this is the tool it calls.
It is the shortest path from “make me a dashboard” to that dashboard being on a wall.
What it does
It wraps the page for unattended signage. A screen has no mouse and no keyboard, so anything that needs a click never advances. The wrapper rotates tab panels and carousel slides on its own, which is the difference between a page that works on a laptop and one that works on a wall.
It completes a fragment. If the assistant hands it a fragment rather than a whole document, it is wrapped into a complete page.
It creates an instance. The app appears in your Content library ready to add to a playlist, rather than existing only as an app definition.
Deploys it.
Updating it later
Publish again with the same name and it updates the app you published before, rather than creating a second one.
That works because the tool remembers the app id by name on the machine it published from. Two consequences worth knowing:
From a different machine that memory is not there, so pass the app_id explicitly to target a specific app.
Prefer passing app_id whenever you know it. It is unambiguous, and it does not depend on the name matching exactly.
Where it fits
Then it is ready for a playlist. Add the instance to a playlist, give it a duration, and assign the playlist to screens. Nothing about it is special once it exists.
What to be careful about
This tool is annotated destructive, because publishing again over a remembered name replaces what was there.
The page is also, at that point, on screens. Anything it fetches is fetched from the device, and anything hard-coded in it is visible to anyone who can read the app. Credentials belong in secrets, not in the HTML.