Anthropic 工程师用 Claude Fable 5.2 协调 1000+ 子智能体写出可启动操作系统原型
Anthropic 工程师拿 Claude Fable 5.2 干了件硬核事：1000 多个子智能体连跑 8 天，写出一个能开机、有内核和窗口系统的 OS 原型。做智能体编排的人应该看看这个案例，比一般 coding demo 能说明规模上限在哪。不过消息来自推特，先当参考。
据 Anthropic 方面的说法，一位 Anthropic 工程师让 Claude Fable 5.2 从空文件夹开始构建一个操作系统。该模型协调了超过 1,000 个子智能体（subagents），历时 8 天不间断地进行代码编写、测试与审查。最终产出一个可在笔记本上启动的工作原型。
这个原型包含自有内核、磁盘/键盘/网络的驱动、持久化文件存储、窗口系统、登录界面、安装程序，并内置了 Claude 功能。技术路径上，它展示了用主模型拆分任务、调度大量子智能体并行处理大型工程项目的编排能力。
需要注意的是，消息源头是一条社交媒体帖子，Anthropic 官方尚未发布完整报告，操作系统级任务对模型长期规划与错误恢复能力要求极高，这一结果是否可复现仍待验证。
This is insane: An Anthropic engineer asked Claude Fable 5.2 to build an operating system from an empty folder.
The model coordinated over 1,000 subagents that spent eight days writing, testing and reviewing code around the clock, according to Anthropic.
The result was a working prototype that boots on a laptop. It has its own kernel, drivers for the disk, keyboard and network, persistent file storage, windows, a login screen, an installer and built-in Claude functionality.
Fable 5.2 looks like the holy grail and it was worth the wait.
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Claude Fable 5.2, Anthropic, 智能体, 编程助手.
Source attribution: curated from kimmonismus (@kimmonismus).
