Remote Control fails to connect with "connect timed out" — reproducible, session confirmed alive

Remote Control repeatedly fails to connect for a session that is demonstrably alive and responsive at the time of the attempt.

Environment:

• Claude Code version: 2.1.168
• OS: Windows 11 Pro, build 26200 (10.0.26200.0), 64-bit

Reproduced on two separate days (2026-09-18 and 2026-09-19) with the same session ("DCS-reports repo access"), same failure both times:

Remote Control failed to connect: Remote Control connect timed out

Details:

• The local Claude Code desktop app was open and the session was actively responding to messages at the exact time of each failed connection attempt — this is not a case of the local host being offline or asleep.
• Toggling Remote Control off then back on for the session reproduces the same timeout both times, rather than a stuck one-off.
• Separately, a browser tab at claude.ai/code connected to this same session showed a "Remote Control disconnected — Claude Code on the computer running this session is offline" banner, even though the session was live on the same laptop the browser tab was open on, and had no other local sessions active to conflict with it.

This points at the Remote Control connection/relay service rather than anything client-side (network, app state, or which machine/session is involved), since the same failure reproduced identically across two days with a confirmed-live session.

Possibly related: #95491 ("Remote Control is not restored for sessions killed by a stealth auto-update") describes a similar end symptom (Remote Control silently unreachable), but I checked the local app logs directly and confirmed no auto-update/relaunch event occurred near either of my two failures — the session was never restarted. So this looks like a distinct mechanism from #95491, not a duplicate, though the maintainers may want to check whether they share an underlying cause.