Claude Code Now Lets Terminal Sessions Talk to Each Other on macOS

Anthropic has fundamentally changed how developers interact with its AI on macOS by introducing cross-session messaging in Claude Code. This new capability eliminates the tedious process of manually copying and pasting context between separate Terminal windows, allowing active coding sessions to communicate directly with one another.

Shipped in Claude Code version 2.1.224 for macOS and Linux, the feature relies on two new underlying tools. The ListAgents tool scans the Mac for other active sessions, while SendMessage delivers the payload. When a session receives a message, it appears as a labeled card containing a link back to the sender. To maintain security and prevent unintended actions, Anthropic's documentation specifies that only text summaries are passed between sessions - conversation history, files, and permissions are strictly excluded. Furthermore, an incoming message cannot autonomously approve permission requests or alter a session's settings.

New in Claude Code: your sessions can now message each other. Instead of having to re-explain yourself in another session, you can now tell Claude to do it. It sends a summary (not your history or files), and the other session picks it up mid-task.

- ClaudeDevs

Unlike Claude's existing subagents and Agent Teams, this feature connects independent sessions that a user is already running locally. Messages exchanged between sessions on the same Mac remain entirely local and never ping Anthropic's servers. For cross-machine communication, messaging is restricted to replies via Remote Control, which links a Mac terminal to Claude's iOS and Android apps. Notably, this feature is currently unavailable on Windows.

The version 2.1.224 release also includes 31 other changes. For Team and Enterprise users, Anthropic introduced a new command to streamline deployments:

claude self-hosted-runner

• 1Password Integration: Claude can now access credentials stored in 1Password to complete browser tasks without exposing the actual passwords to Anthropic's systems.
• iOS Simulator Support: Available in public beta, Claude Code can now build, run, and test iOS apps directly within Apple's Simulator, iterating on the project live.
• Voice Mode Expansion: Claude's voice mode now supports the more powerful Opus and Sonnet models, allowing users to switch models mid-conversation instead of being limited to the Haiku model.

By enabling cross-session messaging, Anthropic is transforming Claude Code from a siloed terminal assistant into a unified, system-aware developer tool. The strict limitation to text-only payloads is a smart security guardrail, ensuring that one rogue session cannot execute privileged commands in another.

However, the real strategic advantage here is the deep integration with macOS and Linux environments while explicitly leaving Windows behind for now. Combined with the new iOS Simulator capabilities and 1Password integration, Anthropic is clearly positioning Claude as the native AI orchestration layer for Apple developers, significantly reducing the friction of multi-agent workflows.