Cloudflare представила @cloudflare/computer для AI-агентов

## A shared filesystem across isolates and containers The @cloudflare/computer package starts with a simple premise: what if we give an agent a primed filesystem, declaratively defined, containing everything required for the task at hand and a selection of execution environments to operate on those files, each with their own pros and cons regarding speed, capability and cost? It turns out that agents today are surprisingly capable of selecting the right environment for the task at hand. A job that only needs to manipulate files, process data, or manage a git repository can run inside an isolate. A command that needs Linux, `npm`, or a native binary can run inside a container. Both work against the same files that are kept in sync with the source filesystem. `npm` BLOG-3464_image2.png [...] `npm` BLOG-3464_image2.png The @cloudflare/computer package provides a durable filesystem that you can use with git repositories, storage buckets or any files you choose. It provides tools that

# '@cloudflare/computer' has been introduced, assigning a dedicated computer to each AI agent, allowing file editing and command execution to be performed within a single work environment. Manus has released 'My Computer,' a desktop application that directly connects an AI agent with files, tools, and applications in the user's local environment. OpenAI has announced an update to its 'Agents SDK' that enables the creation of AI agents that can run for extended periods. GitHub Copilot evolves from an 'AI assistant' to an 'AI development team,' revealing the full scope of its dedicated app. [...] Perplexity Announces 'Personal Computer,' an AI Agent for Macs that Automates PC Operations Like OpenClaw, and Also Offers a Version of 'Perplexity Computer' Specialized for Enterprises Cloudflare announces JAMstack platform 'Cloudflare Pages' that allows you to instantly build a website from source code Microsoft has announced 'Scout,' an always-on AI agent based on Ope

`packages/dofs` (`@cloudflare/dofs`) — Durable Object SQLite-backed virtual filesystem, sync protocol building blocks, and a `@platformatic/vfs` provider for Node. `packages/rpc` (`@cloudflare/computer-rpc`) — capnweb wire types and server/client helpers shared between the Durable Object and `computerd`. `packages/computerd` (`@cloudflare/computerd`) — the `computerd` daemon: a FUSE mount plus HTTP/WebSocket RPC server that runs inside the sandbox container. `packages/computer` (`@cloudflare/computer`) — the top-level Computer package consumed by Durable Objects. Work in progress. `packages/computer-computerd-linux-x64` (`@cloudflare/computer-computerd-linux-x64`) — the prebuilt `computerd` binary for linux-x64, distributed for use in container images. ## Performance [...] Container projects the SQLite state into a sandbox container as a real FUSE mount. A sandbox-side daemon (`computerd`) mounts the state as a filesystem and syncs changes back over a capnweb RPC channel. Full Lin

logo #### 所有版块 # Cloudflare 发布 @cloudflare/computer 给每个 AI Agent 配一台「虚拟电脑」 cryptotalk Cloudflare 发布 cloudflare/computer 给每个 AI Agent 配一台「虚拟电脑」 这台「电脑」是一份存在云端的文件系统，外加几种能操作这些文件的执行环境。 Agent 每敲一条命令，自己决定派到哪个环境去跑：动文件、处理数据、管 git 仓库，在一个毫秒级启动的轻量环境里完成 真需要 Linux、要装包、要跑二进制程序，才去开容器。 Godaddy xhj003 xhj003 #### 你好啊，陌生人! #### 快捷功能区 #### 所有版块 #### 📈用户数目📈 #### 🎉欢迎新用户🎉 闫笑 max-A linjiajia carl12 Leokay mangoqwq huaerye23 kulakulaka

# Thread 49 views ethansavemoneydairy's profile picture ethansavemoneydairy AI Threads 另一個我覺得容易被忽略的是 Filesystem 的設計。 Cloudflare 並不是每個任務都開一個 Container。 如果只是修改 Markdown、編輯程式碼、Git Commit，就使用更便宜的 Isolate Filesystem。 真的需要 npm、Python 或完整 Linux 環境時，才切換到 Container。 代表未來 Agent 不只是會完成工作，也開始會思考如何降低執行成本。 Translate 1 Log in or sign up for ThreadsSee what people are talking about and join the conversation. Log in with username instead

# 【Git Worktree 實戰】指揮 AI Agents 同時運作，搭配 Agent Skills 建立自動化協作體系 ## 工程師下班有約 43000 subscribers 642 likes ### Description 20789 views Posted: 24 Feb 2026 ❝AI 寫程式很快，但「人」才是瓶頸❞ 公司導入 AI 加速開發後，你的團隊是不是也遇到這些問題？ 👉 Code Review 數量暴增，根本看不完 👉 一個 PR 塞了多個功能、上千行程式碼 👉 Commit 訊息隨便寫，出事無法追溯設計邏輯 這支影片將透過實戰，帶你用 Git Worktree + Agent Skills 解決以上實務痛點： ① 沒有版控概念，AI 的快只會成為災難起點 ② 建立 Git Worktree 環境：Git + pnpm + Extensions ③ 用 Agent Skill 自動建立分支與開發規格 ④ 在 Agent Manager 指揮多個 AI Agents 同時執行 ⑤ 用 AI 將雜亂變更整理成乾淨、可追溯的 Commit ⑥ 生成統一格式的 PR 描述，減少 Code Review 負擔 AI 產生的程式碼很多，不該成為不檢查的藉口，我們應該去想怎麼利用 AI 去優化檢查流程 別忘記，當你對 AI 產出內容按下同意的那一刻，所有的錯誤必須由人類來承擔 🫠 📌 參考資源： 影片指令可以參考部落格: 影片 GitHub 專案: Agent Skills 入門影片: Antigravity 入門影片: Cursor 入門影片: 🧑‍💻作者資訊： 林鼎淵的 FB: 林鼎淵的 Medium: 《工程師下班有約》書籍資訊: [...] 🧑‍💻作者資訊： 林鼎淵的 FB: 林鼎淵的 Medium: 《工程師下班有約》書籍資訊: #GitWorktree #AgentSkills #AIAgent #CodeReview #vibecoding #版本控制 #AI開發 #Workflow #工程師下班有約 79 comments ### Transcript: