disableRemoteControl:true in settings.json ignored on v2.1.222; app auto-drafts suggested reply into input box

Status Open
Reported on v2.1.222
Maintainer reply None cached
Activity 0 comments · opened Sep 6, 2026

Summary

Setting "disableRemoteControl": true in ~/.claude/settings.json is not honored on Claude Code v2.1.222. A session started with that setting present is still reachable and steerable from the Claude app's Code / Remote Control view, and the app auto-drafts a "suggested reply" that appears in the session's input box.

Environment

  • Claude Code: 2.1.222
  • Headless CLI sessions running in tmux on Linux (long-lived, non-interactive operators drive them over a chat bridge).
  • Personal (non-Team/Enterprise) account — the org-level toggle at claude.ai/admin-settings/claude-code is unavailable ("You don't have access to organization settings").

Expected

With "disableRemoteControl": true set in user settings before the session starts, the session should not register for Remote Control and should not be steerable / auto-drafted-into from the app.

Actual

  • The setting is present (verified across multiple sessions) yet the sessions still appear online in the app's Code tab and can be controlled.
  • The app populates the session's input box with an AI-generated "suggested reply" (a plausible answer to the session's last message) that the user did not type.
  • Because that text lands in the real input line, it collides with programmatic input to the pane: automated slash commands (e.g. /clear, /compact) injected into the box get concatenated with the phantom text, so the command fails or, worse, the foreign text can be submitted to the model.

Impact

For automation that drives a Claude Code session's input box, an un-suppressable Remote Control channel that writes a suggested reply into that box is a correctness and safety problem: injected commands are corrupted, and un-vetted suggested text risks being submitted.

Notes / questions

  1. Is disableRemoteControl the correct key, and is it expected to take effect at session start? (remoteControlAtStartup is also referenced in some docs.)
  2. Is there a documented way to disable Remote Control for a personal account (no org admin panel)?
  3. Is the "suggested reply auto-draft into the input box" an intended feature? It does not appear to be documented.

View original on GitHub ↗