Claude Code Remote Control: Mobile & Web Bridge
Claude Code Remote Control separates the agent’s control surface from its execution environment. You kick off a task in a local terminal, then monitor and steer it from a phone or browser — without moving the workload to the cloud. Unlike a cloud IDE, local context (filesystem, environment variables, and MCP servers) stays live on the host and reachable to the agent.
Architecture and security
The bridge is designed so the host never exposes itself to the network.
Outbound-only polling. The local machine opens no inbound ports. The claude process makes an outbound HTTPS connection to Anthropic’s API and polls for instructions. Only chat messages and tool outputs cross the encrypted bridge — files and MCP servers stay local, and the mobile app is just a remote window into the local session.
Session persistence. Reconnection is automatic. If the host sleeps or the network drops, session state is preserved; when the machine comes back online, the bridge re-establishes itself.
Setup and usage
Requirements
- Plan: Claude Pro or Claude Max.
- Version: Claude Code v2.1.52 or higher.
- Auth: authenticated via
/login.
Commands
| Command | Function |
|---|---|
claude remote-control |
Starts a new remote session and generates a QR code / URL. |
/rc |
Toggles Remote Control from within an active session. |
/config |
Set Enable Remote Control for all sessions to true. |
Why it matters
Remote Control enables a “headless” style of agent management: long-running tasks — refactoring a codebase, running a test suite — are launched at a workstation but supervised and guided from a mobile device while they run. It’s the mobility layer for the parallel, fleet-commander way of working.
