[BUG]

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Here’s a comprehensive bug report you can copy and paste:

Title: Remote Control sessions silently drop and cannot recover on iOS app and mobile Safari — messages appear to send but are never received
Environment
• Claude Code version: 2.1.104
• Host OS: Ubuntu 24.04.4 LTS (DigitalOcean droplet, not local machine)
• Remote clients: Claude iOS app + mobile Safari (claude.ai/code)
• Plan: Claude Max
• Session type: Remote control sessions initiated via claude CLI on droplet
Description
Remote control sessions consistently drop their connection and become completely unresponsive on both the iOS app and mobile Safari. This has persisted for over 12 hours across reboots of both the droplet and the iOS app.
Symptoms
• Sessions appear connected and “Interactive” in both the iOS app and Safari
• Messages appear to send (no error shown, message appears in conversation)
• Spinner runs indefinitely — agent never receives the message
• Occasionally shows “Message could not be sent. An unknown network error has occurred”
• Force-quitting and reopening the iOS app does not fix it
• Rebooting the droplet does not fix it
• Safari (claude.ai/code) exhibits identical behavior to the iOS app — not a client-specific bug
• Two sessions were auto-archived with no user action
• The droplet itself is healthy — tmux sessions active, node processes running, Claude Code responding normally in Termius
What works
• Typing directly in Termius (SSH → tmux attach) works reliably
• The agent continues running on the droplet even when the app/Safari connection is dead
What doesn’t work
• iOS app remote control sessions (persistent)
• Mobile Safari claude.ai/code remote control sessions (persistent)
• Disconnect/reconnect via /remote-control — cannot reach the prompt because the session appears frozen
Steps to reproduce

1. Start Claude Code on a remote Linux server (DigitalOcean droplet)
2. Connect via iOS app remote control session
3. Leave phone idle or switch apps briefly
4. Return to session — appears connected but messages never reach the agent
5. Try Safari (claude.ai/code) as alternative — same behavior

Impact
This makes remote control sessions completely unusable for the intended use case — managing long-running agent tasks from a phone without physical access to the terminal. The only workaround is SSH + tmux, which defeats the purpose of the feature.
Related issues
• #34255 — automatic reconnection doesn’t work
• #28571 — session fails to resync after connection drop
• #29726 — iOS app drops connection on background/foreground
• #41670 — iOS app fails while browser succeeds (same “unknown network error”)
Additional notes
This appears to be a server-side WebSocket/streaming connection issue, not a client-side bug, since both the iOS app and Safari exhibit identical failure modes. The issue persisted across a full droplet reboot, iOS app reinstall attempt, and multiple reconnection attempts over 12+ hours.

File it at: github.com/anthropics/claude-code/issues/new

What Should Happen?

Here’s a comprehensive bug report you can copy and paste:

Title: Remote Control sessions silently drop and cannot recover on iOS app and mobile Safari — messages appear to send but are never received
Environment
• Claude Code version: 2.1.104
• Host OS: Ubuntu 24.04.4 LTS (DigitalOcean droplet, not local machine)
• Remote clients: Claude iOS app + mobile Safari (claude.ai/code)
• Plan: Claude Max
• Session type: Remote control sessions initiated via claude CLI on droplet
Description
Remote control sessions consistently drop their connection and become completely unresponsive on both the iOS app and mobile Safari. This has persisted for over 12 hours across reboots of both the droplet and the iOS app.
Symptoms
• Sessions appear connected and “Interactive” in both the iOS app and Safari
• Messages appear to send (no error shown, message appears in conversation)
• Spinner runs indefinitely — agent never receives the message
• Occasionally shows “Message could not be sent. An unknown network error has occurred”
• Force-quitting and reopening the iOS app does not fix it
• Rebooting the droplet does not fix it
• Safari (claude.ai/code) exhibits identical behavior to the iOS app — not a client-specific bug
• Two sessions were auto-archived with no user action
• The droplet itself is healthy — tmux sessions active, node processes running, Claude Code responding normally in Termius
What works
• Typing directly in Termius (SSH → tmux attach) works reliably
• The agent continues running on the droplet even when the app/Safari connection is dead
What doesn’t work
• iOS app remote control sessions (persistent)
• Mobile Safari claude.ai/code remote control sessions (persistent)
• Disconnect/reconnect via /remote-control — cannot reach the prompt because the session appears frozen
Steps to reproduce

1. Start Claude Code on a remote Linux server (DigitalOcean droplet)
2. Connect via iOS app remote control session
3. Leave phone idle or switch apps briefly
4. Return to session — appears connected but messages never reach the agent
5. Try Safari (claude.ai/code) as alternative — same behavior

Impact
This makes remote control sessions completely unusable for the intended use case — managing long-running agent tasks from a phone without physical access to the terminal. The only workaround is SSH + tmux, which defeats the purpose of the feature.
Related issues
• #34255 — automatic reconnection doesn’t work
• #28571 — session fails to resync after connection drop
• #29726 — iOS app drops connection on background/foreground
• #41670 — iOS app fails while browser succeeds (same “unknown network error”)
Additional notes
This appears to be a server-side WebSocket/streaming connection issue, not a client-side bug, since both the iOS app and Safari exhibit identical failure modes. The issue persisted across a full droplet reboot, iOS app reinstall attempt, and multiple reconnection attempts over 12+ hours.

File it at: github.com/anthropics/claude-code/issues/new

Error Messages/Logs

Steps to Reproduce

Here’s a comprehensive bug report you can copy and paste:

Title: Remote Control sessions silently drop and cannot recover on iOS app and mobile Safari — messages appear to send but are never received
Environment
• Claude Code version: 2.1.104
• Host OS: Ubuntu 24.04.4 LTS (DigitalOcean droplet, not local machine)
• Remote clients: Claude iOS app + mobile Safari (claude.ai/code)
• Plan: Claude Max
• Session type: Remote control sessions initiated via claude CLI on droplet
Description
Remote control sessions consistently drop their connection and become completely unresponsive on both the iOS app and mobile Safari. This has persisted for over 12 hours across reboots of both the droplet and the iOS app.
Symptoms
• Sessions appear connected and “Interactive” in both the iOS app and Safari
• Messages appear to send (no error shown, message appears in conversation)
• Spinner runs indefinitely — agent never receives the message
• Occasionally shows “Message could not be sent. An unknown network error has occurred”
• Force-quitting and reopening the iOS app does not fix it
• Rebooting the droplet does not fix it
• Safari (claude.ai/code) exhibits identical behavior to the iOS app — not a client-specific bug
• Two sessions were auto-archived with no user action
• The droplet itself is healthy — tmux sessions active, node processes running, Claude Code responding normally in Termius
What works
• Typing directly in Termius (SSH → tmux attach) works reliably
• The agent continues running on the droplet even when the app/Safari connection is dead
What doesn’t work
• iOS app remote control sessions (persistent)
• Mobile Safari claude.ai/code remote control sessions (persistent)
• Disconnect/reconnect via /remote-control — cannot reach the prompt because the session appears frozen
Steps to reproduce

1. Start Claude Code on a remote Linux server (DigitalOcean droplet)
2. Connect via iOS app remote control session
3. Leave phone idle or switch apps briefly
4. Return to session — appears connected but messages never reach the agent
5. Try Safari (claude.ai/code) as alternative — same behavior

Impact
This makes remote control sessions completely unusable for the intended use case — managing long-running agent tasks from a phone without physical access to the terminal. The only workaround is SSH + tmux, which defeats the purpose of the feature.
Related issues
• #34255 — automatic reconnection doesn’t work
• #28571 — session fails to resync after connection drop
• #29726 — iOS app drops connection on background/foreground
• #41670 — iOS app fails while browser succeeds (same “unknown network error”)
Additional notes
This appears to be a server-side WebSocket/streaming connection issue, not a client-side bug, since both the iOS app and Safari exhibit identical failure modes. The issue persisted across a full droplet reboot, iOS app reinstall attempt, and multiple reconnection attempts over 12+ hours.

File it at: github.com/anthropics/claude-code/issues/new

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.104

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_