[feature] LLM - Additional Re-runnable question/prompt on the actual extracted text - each time

Would be pretty handy if in the "history" page, there could be a tab with some prompt you could run on each change

which could also be outputted as a token? unsure, or in 'edit'

for example, the page changes, and its a whole slab of text, but i want a general report like "give me a summary of the events"

this is additional to just the CHANGE summary, i also want some way to get a general query on the whole text document

Start with the history page and the existing CHANGE summary flow. Determine how a prompt should run against the full extracted text for each change, and clarify whether its result belongs in a token or in edit mode. Done means the issue’s example query can be rerun on each change and its output is available as specified.

Written by the indexing model from the issue text.