Your filesystem, tools, MCP servers, and project configuration stay on that machine.
Read Anthropic’s guide ↗Why is Claude Remote Control not working?
Check the six boundaries Anthropic currently documents. Get the local or organization fix first. Hosting appears only when Remote Control is ready but the machine cannot stay online.
Describe the environment that starts Claude.
Select “not sure” when needed. The report will give you a verification command rather than pretending the check passed.
Remote Control is usually the free answer.
Run server mode inside tmux or screen. Do not buy a second host for the same job.
Generate the tmux setup →Compare Terminalbox only when the required process must execute somewhere else.
Ask for an honest fit check →The failures this checker separates.
Why does Remote Control say it requires a claude.ai subscription?
Remote Control needs an eligible full claude.ai login. API-key-only authentication is not supported. Open Claude Code and use /login with the claude.ai account you use on mobile or web.
Why does a full-scope login token matter?
Long-lived tokens created with setup-token support inference but not the Remote Control session scope. Sign in again through /login.
Can I use Remote Control through Bedrock, Google Cloud, Foundry, or an LLM gateway?
Anthropic currently says Remote Control is unavailable on those cloud-provider paths and is disabled when ANTHROPIC_BASE_URL points away from api.anthropic.com.
Which environment variables can disable Remote Control?
DISABLE_TELEMETRY, DO_NOT_TRACK, CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC, and DISABLE_GROWTHBOOK disable the feature-flag evaluation Remote Control eligibility depends on.
Does Remote Control keep working while my laptop is asleep?
It reconnects when the machine returns, but execution remains on that machine. If the local Claude process stops, the session ends. Use an existing always-on server or a persistent host only when work must execute while the laptop is off.
Does this checker upload my Claude settings?
No. Your selections and generated commands stay in this browser tab. Terminalbox records only a random per-tab session and bounded labels such as tool view, diagnosis outcome, and clicked destination.