Anthropic于9月17日启动Claude Code Projects的公开测试,旨在通过引入新的协调机制和工具来优化开发者的并行工作体验 1。此次更新的核心功能包括主动协调器(proactive coordinator)、只读线程分享、定时提醒以及/btw和/fork斜杠命令 1。这些新功能意在减少开发者同时监控多个云端会话的工作负担 1。目前,该公测版仅向Pro和Max订阅用户开放,Team和Enterprise计划的用户暂时无法使用此功能 1。
在资源与安全方面,所有并行运行的工作线程将共用用户的Claude Code使用额度,且每日新线程创建数量限制为200个 1。出于安全考虑,涉及本地计算机访问权限的线程禁止通过链接进行分享,以防止本地文件内容泄露 1。对于新项目,系统默认配置协调器和工人线程均使用Opus模型,其中协调器处于低努力模式,而工人线程则运行在高努力模式下 1。
Anthropic has introduced a public beta for updated Claude Code Projects, enabling developers to run parallel AI workers with minimal manual oversight. Launched on September 17, this update aims to reduce the workload associated with monitoring multiple cloud sessions by introducing a proactive coordinator and new slash commands such as /btw and /fork 1. The feature is currently accessible only to Pro and Max subscription users, while Team and Enterprise plans do not yet support this functionality 1.
The update includes several core enhancements designed to streamline collaborative workflows, including read-only thread sharing and scheduled reminders 1. All parallel worker threads consume from the user’s single Claude Code usage quota, with a hard limit of 200 new threads created per day 1. To maintain security, threads that require access to the local computer are restricted from being shared via links, preventing potential leakage of local file contents 1. By default, new projects utilize the Opus model for both the coordinator and worker threads, configuring the coordinator in low-effort mode and workers in high-effort mode 1.
评论
还没有评论,欢迎留下第一条。