Ah… that’s kind of frustrating? It’s always felt like there’s a bit of a shadow data model that the UI doesn’t expose, I guess that’s it! Seems related to some of my earlier feedback about the confirm and cleanup button having different functions.
You’re definitely right about all the confirmations coming across, no issues there. Does the model look at the individual point data or just the confirmations? I’d still like it to be clean for export purposes, but could be convinced to not care as much I suppose.
Are the classifier tasks that you think will consolidate these represented in the debug tab? Doesn’t seem like anything is in the queue:
Finally, RE Claude, that makes sense! I’m slightly disappointed you don’t have some sort of crazy context management workflow (other than pair programming), and I’m happy you disclose where it’s primarily used with the account.
Thanks again for your time.