Anthropic launches Claude Code cross-session communication function to improve multi-task collaboration efficiency. Artificial intelligence startup Anthropic today launched a major update for its AI programming tool Claude Code.

Anthropic launches Claude Code cross-session communication function to improve multi-task collaboration efficiency. Artificial intelligence startup Anthropic today launched a major update for its AI programming tool Claude Code. In the latest v2.1.224 version, Claude Code running on macOS and Linux platforms officially supports cross-session messaging, allowing different terminal sessions to send information to each other, share findings, answer questions, and coordinate work. For technicians who are used to running multiple Claude Code instances at the same time to work on different modules of the same project, this feature greatly simplifies the complexity of multi-tasking collaboration. When a technical change in one session has a disruptive impact on other modules, Claude will send an early warning to the affected session in advance; and when a session solves a key problem that other tasks are getting stuck on, relevant answers can also be transmitted across sessions in a timely manner. According to the official description, typical application scenarios of this feature include submitting analysis conclusions, coordinating parallel work trees (worktrees), obtaining the latest status of long-running tasks, and achieving cross-machine response communication. In terms of operating experience, users do not need to spend time conceiving a specific message structure. They only need to provide the core communication intention, and Claude can automatically generate a complete text message. It is important to note that the cross-session messaging feature maintains strict security restrictions on permission management. This mechanism cannot be used to approve permission requests or change global configurations, and commands like /compact are transmitted as plain text rather than executed directly. If the received conversation needs to perform specific operations based on the message content, it will still proactively request permission approval from the user. via cnBeta.COM - Chinese industry information station (author: source: cnBeta.COM)