I've seen first-person/third-person switches done two ways. First, including an italicized first-person narrative, of varying length, at the beginning of every chapter. Brandon Sanderson's
Mistborn did that, and so did Timothy Zahn in his
Star Wars: Thrawn. I can't remember any other examples of that off the top of my head, but I'm sure there are others. In some versions of this, the person speaking isn't even necessarily known by the reader.
The second way is something I've seen far less often, and only in one book that I can think of right now. Connie Willis'
Doomsday Book has multiple short(ish) interim chapters of Kivrin Engle's journal/log, right beside her third-person narrative about occasionally the same events, with entire sections within her POV chapters that go from one to the other, both first-person narrative from Kivrin through her journal, and third-person narrative showing her experiencing further events that didn't necessarily make it into the log.
In all three cases, there was an acknowledged in-universe text (or recording) in which the character involved is actually known to have laid down their thoughts for others to see. So our third-person narrative is simply showing us a particular character's in-universe words, however long or complicated or implausibly well-crafted they might be--and in so doing, the book is letting itself become a first-person narrative for the duration.
@msstice's example, I imagine, would
require some kind of differentiation between the first and third person, otherwise your interim sentences, such as "Only one bullet left"--which, according to reader interpretation, could belong to either first
or third--will be left existing in a kind of gray area where it's super unclear which one we're currently in. The most helpful thing for it is going to be to make sure the transitions between first and third person are crystal clear for the reader. That's why people tend to italicize things like
I have to make it count, and call it a direct thought. It's instantly recognizable and therefore doesn't trip up the reader. That being said, there may be plenty of other ways you can indicate the same thing
without italicizing the thought, such as paragraph changes or quotes or using an obviously different language/dialect, but I've never looked into that and I couldn't tell you the best one.
Thinking back, I'm pretty sure Orson Scott Card's Enderverse books have made use of unitalicized "I" references, but they were still obviously direct thoughts, and it was always either signaled well in advance by the text or simply paired with actual dialogue tags along the lines of "What? he thought. No. I won't do that." or "What did you think I meant, he didn't say. You
knew I was in danger." A little nonstandard, of course, but still perfectly understandable within the text.
So that would be my bottom line. As long as the reader can clearly understand
when and
where the shift happens, plus who is speaking and to whom each "I" and "you" refers, then one can get away with a surprising amount of switching in the narrative. It'll definitely be nonstandard, but it will at least
work in your story. I've found that in many cases, things that are definitely mistakes when done accidentally are "stylistic choices" when done deliberately and with genuine craft and care. Put careful thought into what you're doing, and then do it with confidence--and then the world, as Terry Pratchett has said, is your mollusc.
