I would like the ability to inject images. It would actually inject the enter <mediaobject> tag, with the <textobject><phrase> (the alternate text) being populated by the image description. This injection could be wrapped in a <figure> if we needed to give the image a title / caption. Presumably syntax would be something like:
<!-- InjectImage: 28.svg -->
Or just <!-- InjectImage: 28 -->
The second way is less prone to error as Skynet would figure out the image type from its own metadata.
The Image view now lists the XML that is used to display an image, and the docbook xml is also exposed as a template in the new UI.