/remote-control client falls behind on long-running sessions, doesn't recover after reopen
Resolved 💬 3 comments Opened May 10, 2026 by goodgoblin76 Closed May 14, 2026
Summary
When activating /remote-control on a long-running CLI session, secondary clients struggle to stay in sync — they get stuck on older messages and don't backfill or catch up to the live state. Closing and reopening the lagging client does not fix it.
Steps to reproduce
- Run a long Claude Code session in the terminal (many messages of history).
- Run
/remote-controland connect a phone or browser client — works fine. - From a second device (e.g. another phone, or browser), connect to the same session via the session list on claude.ai/code or the mobile app.
- The second client lands on the correct session but is stuck several messages behind the live state.
Expected behavior
Per the docs, all connected clients should stay in real-time sync, and a newly-connected client should backfill the full conversation history.
Actual behavior
- Second client shows older messages only; new messages from the live session don't arrive.
- Closing and reopening the app on the lagging client does not resolve it.
- Same session confirmed (not two different sessions).
- Pattern is reproducible on long-history sessions specifically; fresh sessions sync fine.
Environment
- macOS (Darwin 25.2.0)
- Claude Code CLI, recent version
- Lagging client: iPhone Claude app
- Primary client: terminal CLI on the Mac
Notes
Suspect the long-session backfill path is the culprit — fresh sessions don't exhibit this. Workarounds attempted: close+reopen the lagging app (no effect), confirm same session ID (yes).
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗