Users are advocating for a built-in query replace feature in Discourse’s editor, similar to GitHub’s web-based editor, to avoid manually copying text to external editors for bulk replacements jidanni. merefield and Arkshine support the idea, with Arkshine sharing a visual reference of GitHub’s interface. Lilly demonstrates the functionality using an AI popup to replace text (e.g., “words” → “things”), showing before/after results. Arkshine suggests allowing users to choose or create custom prompts. Falco notes this should be considered for the new composer experiment, and sam humorously illustrates LLM-powered replacements (e.g., “animals” → “amphibian animals”), highlighting the potential of AI integration. The feature would streamline editing, especially for long-form content.
Let’s say you have a lot of "33"s that you want to change into "44"s.
Well on GitHub when we’re editing source code they’ve got a nice query replace function right there in the web based editor.
So there’s “no reason” why Discourse couldn’t have the same! (Worldwide.)
That way we wouldn’t need to copy our text into some other editor, do the replacements, and then paste it back in.
Yes please!
With Discourse long-form writing this has become more and more an issue for me.
It’s currently necessary to move it out to a third party editor (manually), make the changes and then put them back. Ugh!
Excellent Contribute > Feature request!
OK this works. using prompt replace all "words" with "things" in the AI popup shown in the screenshot above when you select the text:
before:
Here is a big block of text with lots of repeating words. All the words are words that are words and more words. A group of donkeys are always words, but a group of words are not always donkeys.
after:
Here is a big block of text with lots of repeating things. All the things are things that are things and more things. A group of donkeys are always things, but a group of things are not always donkeys.
It would be helpful and interesting to have the ability to choose prompts from a list. 
(even creating your own)
Something to consider on the new composer experiment @renato
“replace all animals with similar amphibian animals”
Here is a big block of text with lots of repeating words. All the words are words that are words and more words. A group of salamanders are always words, but a group of words are not always salamanders.
LLMs sure can be useful.
