Discourse’s backups are SQL, but I needed a Markdown or HTML export. I asked ChatGPT to write a quick Ruby script that loops over the possible post IDs and downloads the Markdown using the special “/raw” URLs in Discourse. Reviewed it line by line and tweaked it slightly, but it was essentially correct on the first try. It took minutes from conception to running.
Maybe there was an existing Discourse solution for this. But the fact that in the time it would take to find and install another solution, I could have AI write my own custom solution… Still just amazing.
