Hedwig Vollers

As the only one who works on my product at this point in time, I've had a habit of neglecting all sorts of documentation for ages. I was familiar with my product well enough not to get lost in details and to know what was going on and why. But after a two-week break, I couldn't recall some of my own decisions regarding a particular aspect of development.

Now I make myself take a minute or two to leave comments to any decision I've made that wasn't obvious enough, so that I won't have to go through the commit history each time I return to it. It's already helped me save some of my time.

Someone posted a query regarding export timeouts on files over roughly 40MB. No clue whatsoever. I would've previously written a response along the lines of how much I appreciated his patience while I checked it out.

I just admitted that I did not know what the problem was, told him what I wanted to look at, and if it was alright to share the problematic file with me. Four minutes later I received the file and found out it was a 30 second hardcoded timeout I'd written way back in 2024, probably forgot about.

I ve run into this with longer agent runs and I m curious how others handle it.

The original issue might contain the scope, constraints and evidence required. But when another model or session picks up the work later, those conditions often get reduced to a summary.