Background subagents continue after stop, unclear what's running
Status Closed — duplicate
Reported on v2.1.56
Maintainer reply None cached
Activity 3 comments · opened Feb 25, 2026 · closed Mar 1, 2026
Problem
When pressing the stop button during a session that has spawned background subagents:
- The main conversation stops, but subagents continue running and keep prompting for permissions
- The spinner continues with no indication of what's still active
- From the iOS remote control, this is especially confusing — you see permission prompts appearing for work you thought you stopped
Expected behavior
- Stop should cancel all active subagents, or
- The UI should clearly show which subagents are still running with individual stop controls
- Permission prompts from subagents should identify which agent is requesting
Suggested approach
Show an active agent list when subagents are running:
Main session: stopped
├─ Explore agent (a8bf...): still running [Stop]
└─ Research agent (a5b8...): still running [Stop]
On mobile, each running agent could be a dismissible card with its own stop button.
Environment
- Claude Code 2.1.56
- iOS Claude app via remote control
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗