--file flag and ask about it. The two work differently, and which one you get depends on what you attach.
Image Input
Attach a picture and a vision-capable model reads it: describe a screenshot, transcribe a receipt, read a chart.
Documents
Attach a PDF or office file. Nito parses it to text so any model can reason over it.
Images and Documents Take Different Paths
The practical rule: a picture needs a model that can see, while a document works with any model because it becomes text first. If you attach an image to a model that cannot accept images, Nito tells you clearly rather than guessing. See Image Input for that behavior, and Document Parsing for how files are read.
Attach a file with
--file on ask (it is repeatable), in a session, or from the terminal.