Your agent needs you. Where should it find you?
I had Claude Code running behind a notes window. It reached a permission prompt and sat there while I kept working. I was on the same Mac the whole time.
That felt worse than missing a finished task. I was not away. The terminal was simply not where my attention was.
A reply on my first thread here changed how I think about this. @tristan666666 wrote that agent runs should not own your attention. The reply split the problem into two distances: quiet awareness when you are at the desk, and a real interruption when you are away.
I think that split is right. What I cannot settle is the default.
A. Keep everything inside the terminal.
B. Show quiet state in the menu bar or notch, then expand only when I need to act.
C. Use a normal desktop alert.
D. Send it to my phone even when I am still at the Mac.
E. Use every surface and let me silence the ones I hate.
Which one would you actually leave enabled after a month?
I am especially curious about the last permission you missed. Were you away from the computer, or just working in another window?
Disclosure: I build Pushary. We are working on the Mac side of this problem now. I am asking before we lock the default, because notification settings are easy to add and hard to make people revisit.
Replies
Luke's split is the right one, and your follow-up about who assigns severity is the harder half.
In one audit workflow of mine the model that finds a flaw has nowhere to record how serious it is. Severity is a constant of the checklist, fixed before it reads your material, and the finder can only look it up.
It is the only one of mine built that way, so I would not push it as a general rule.
Pushary
@cmumulle That checklist-owns-severity pattern is useful because it prevents the finder from grading its own work. It feels strongest when the workflow is narrow and the rubric is stable. For open-ended tasks, I would let the agent propose severity but require policy to decide the interruption level.
B for me. our agent asks before it posts anything and the hard part is remembering what i already approved, so the audit trail matters more to me than the alert
Pushary
@niveditha_patluri1 That is a useful correction: the alert gets you to the decision, but the audit trail is what prevents the same question from becoming permanent friction. I would want each entry to show what was approved, its scope, and whether it was one-time or reusable. For posting, a reusable rule probably needs tighter boundaries than “this agent can post.” Would you scope it by destination and action, or also by content type?
@aadilghani destination and action i think. content type sounds fine until you actually write the rule, a post and a reply are the same content type but we feel totally different about them.
the thing that made us split it was replies showing up in someones notifications. posting to our own feed nobody has to see
Pushary
That makes sense. The meaningful boundary is not content type, it is who gets pulled into the action. Posting to your own feed can be a reusable permission, while replies and mentions should stay approval-gated because they create a notification for someone else. I would encode that as destination + action + audience, with the audit trail showing which boundary was used.