Request: 'stash' prompt drafts (to write a different one first)

What feature would you like to see?

Use case:

I'm writing a very long prompt. Then I decide I want to use a different model (such as gpt-5-codex-high). I want to "stash" the prompt draft, run /model, and then return to editing my prompt.

Are you interested in implementing this feature?

_No response_

Additional information

_No response_