Remote Control: Frequent disconnections every 20-60 min on v2.1.71 (was stable 24h+ on prior versions)

Description

Remote Control sessions disconnect every 20-60 minutes on v2.1.71 (macOS). The same setup previously maintained stable connections for 24+ hours (including overnight while the machine was idle/asleep) on earlier versions.

The issue started after upgrading to recent versions (v2.1.70/v2.1.71). Each time it disconnects, the session must be fully reconnected — there is no automatic recovery.

Environment

• Claude Code version: 2.1.71
• OS: macOS (Darwin 25.3.0, Apple Silicon)
• Node: v24.14.0 (via nvm, also tested with v20.20.0)
• Connection: Stable home network (no VPN/proxy)
• Plan: Max (Remote Control enabled)
• Config: "Enable Remote Control for all sessions" is ON via /config

Steps to Reproduce

1. Start Claude Code session with remote control enabled (either claude remote-control or /remote-control)
2. Connect from phone/browser via the provided URL
3. Use the session normally
4. Within 20-60 minutes, the remote connection drops
5. Must reconnect manually each time
6. Repeat — happens consistently every session

Expected Behavior

Remote Control should maintain a stable connection indefinitely (as it did on prior versions — confirmed 24+ hour sessions including overnight idle periods).

Actual Behavior

• Connection drops every 20-60 minutes with no warning
• No automatic reconnection
• Must manually reconnect for each session
• Pattern is consistent and reproducible across multiple days

What Changed

This was not an issue before upgrading to v2.1.70+. The connection was rock-solid for 24+ hours continuously on prior versions. The regression appears to coincide with the v2.1.70 release which changed the relay polling interval from frequent (1-2s) to infrequent (~10 min) to reduce server load. Any transient network hiccup during the longer polling gap may now cause the relay to declare the session dead.

Related Issues

• #28571 — Session fails to resync after connection drop
• #28532 — Frequent disconnections requiring page refresh (closed as duplicate)
• #29726 — iOS app drops connection on background/foreground
• #28914 — Mobile app connection dies overnight

Suggested Investigation

1. Is the v2.1.70 polling interval change causing sessions to miss heartbeats during brief network fluctuations?
2. Could the relay implement a more aggressive retry/backoff before declaring a session dead?
3. Is there a way to make the polling interval configurable for users who need persistent connections?