Kinda could just use Lemmy / Piefed like that already if it's text or images.

For documents we'd need a new feature