I just added an LLM prompt enhancer to my workflow that i feed a simple prompt, lets it generate a new one with a fixed seed, and then sends that to conditioning. The fixed seed so that it only creates a new prompt if my initial prompt changes.

However, sometimes i want to edit the generated enhanced prompt because it missed out on some important detail, or i want to remove something else. The problem is that it's hooked up into a text input of a normal prompt node that doesnt allow for editing, and i want to keep the function where it only runs the LLM if my initial prompt changes, even if i decide to edit the generated enhanced prompt. So i would like a setup similar to what i have drawn in pic related:

The LLM gets my prompt, creates a new one and sends it to a mystery node that keeps it stored, and then sends it to a new node that is able to display it, and lets me edit it, before sending it off to condition.

Does anyone have any suggestions how to achieve this?