Claude Code Sessions Can Now Message Each Other, Which Is Stranger Than It Sounds
Two terminal windows running the same AI coding tool can now pass notes back and forth without you copying anything. It is a small feature that quietly changes what a working session is.
Anthropic has added cross-session messaging to Claude Code on macOS and Linux. Two separate sessions, running in two separate terminal windows, can now send each other text. Until now, if one session worked out something the other needed to know, you were the transport layer: you read the answer, copied it, and pasted it into the other window.
It works in both directions. You can ask a session to put a question to another one and get the answer back where you are. Claude can also send a message on its own initiative, for example when a change it just made affects what a parallel session is working on. Anthropic’s documentation lists the expected uses: sharing findings, coordinating work that runs side by side, and checking on the status of something long-running.
Two details are worth knowing before you turn it on. On the same machine, messages stay local. Between different computers, they route through Anthropic’s servers, and in that case only replies are possible, not unsolicited messages. Administrators can switch the whole thing off in settings, which matters for anyone running this in a company. It is also not available if you access Claude through Amazon Bedrock, Google Cloud’s agent platform, or Microsoft Foundry.
The reason this is more interesting than a convenience feature is what it says about how these tools are being used. A year ago, a session was a conversation: you asked, it answered, you closed the window. Increasingly a session is a worker with a job, and people run several at once on different parts of the same project. Once you have several workers, you have a coordination problem, and every coordination problem eventually grows a messaging system. That is what just happened here. The honest caveat is that coordination between semi-autonomous agents is also where confusion compounds, because a wrong conclusion in one session can now travel to another one and be treated as established fact.
What this means for you: if you use Claude Code with more than one window open, this removes a genuinely annoying chore and you can start using it now. Keep an eye on the routing detail: same machine stays local, across machines it passes through Anthropic. If you are in a company with rules about where code context may travel, check that before enabling it, and know the admin switch exists. If you do not write code, the useful signal is the direction of travel. AI tools are quietly acquiring the furniture of teams, with messages, status updates and handovers between them. That is a different thing from a chatbot, and it is arriving without much fanfare.
Sources
Source: https://code.claude.com/docs/en/cross-session-messaging
A Fields Medalist Who Wrote a Paper on AI Extinction Risk Just Joined OpenAI to Work on Safety
Number theorist Jacob Tsimerman is moving from the University of Toronto to OpenAI. His argument is that AI still runs on trial and error, and that mathematicians are the people who could change that.
