One problem I keep hitting: with unattended sessions, or something done a while ago, I find it difficult to reconstruct what I asked Claude Code to do and what it actually ended up doing. Yes, there are sessions I can scroll, and there are jsonl files I can grep. And yes, there are session replays. But it is still quite time consuming. Not to mention going through all the git diffs. And then there is the issue of multiple sessions coordinating their own agents on the same repo.
So I'm working on a CLI tool that helps me reconstruct this interval by interval, and then do a deep dive where needed. So far it has been quite interesting - it is teaching me a lot about how the Claude Code harness works.
If you have a similar problem, I would be very interested to know how you solved it.