Skip to content
Anthropic 讓 Claude Platform 全面提供 Computer use、browser tool、Skills API 與 Files API episode artwork

EPISODE · Aug 21, 2026 · 10 MIN

Anthropic 讓 Claude Platform 全面提供 Computer use、browser tool、Skills API 與 Files API

from EasyVibeCoding Podcast · host ClaudeDevs

Anthropic 讓 Claude Platform 全面提供 Computer use、browser tool、Skills API 與 Files API。 發布重點 ClaudeDevs 於 2026 年 8 月 21 日分享這項更新,核心主張是:即使企業應用程式沒有 API,Claude 仍能透過 Computer use 或 browser tool 執行工作;團隊則可用 Skills API 封裝並版本化既有程序,再透過 Files API 重複使用輸入與輸出檔案。來源文件標示相關功能於 2026 年 8 月 20 日在 Claude Platform 正式全面可用。 這次更新包含四個互相搭配的元件: Computer use:透過 screenshot、滑鼠與鍵盤控制 desktop environment。 browser tool:直接讀取網頁結構並操作元素,減少只依賴螢幕座標的脆弱性。 Skills API:上傳、版本化並重複套用團隊的程序與專業知識。 Files API:檔案上傳一次後,後續 request 可用 file_id 參照,不必重複傳送。 Skills API 與 Files API 是 Claude Managed Agents 的基礎。Computer use 則是由應用程式在自控環境執行的 client toolset,目前不在 Managed Agents 提供;需要操作沒有 API 的系統時,應用程式必須執行工具呼叫,再把結果送回 Claude。Browser tool 適合只在網頁內完成的工作。 Computer use 的多動作執行 新版 computertoolset20260801 改變了 Computer use 的互動方式。Claude 不再每次 model call 只執行一次點擊或輸入,而是能在同一個 turn 依序產生數個動作,例如點擊、輸入、按鍵與截圖。ClaudeDevs 表示,早期存取客戶的每個 task round trip 減少 20–40%,因此工作完成得更快,成本也較低。 computertoolset20260801 是 Anthropic 定義的 client toolset。單一 {"type": "computertoolset20260801"} 會提供 17 個 member tools,包含: screenshot、zoom leftclick、rightclick、middle_click doubleclick、tripleclick leftclickdrag mousemove、leftmousedown、leftmouse_up cursor_position scroll type、key、hold_key、wait Claude 的 tooluse block 會帶有 "toolsetname": "computer",而 member tool 的參數直接放在 input,不再使用舊版的 input.action。應用程式必須依 (toolset_name, name) pair 分派,不能只讀取第一個 tool call,也不能假設同名的 screenshot 或 key 一定來自同一個 toolset。 Computer use 的簡單呼叫方式如下,這個 request 同時宣告 Computer use、text editor 與 bash: `bash curl https://api.anthropic.com/v1/messages \ -H "content-type: application/json" \ -H "x-api-key: $ANTHROPICAPIKEY" \ -H "anthropic-version: 2023-06-01" \ -d '{ "model": "claude-opus-5", "max_tokens": 1024, "tools": [ {"type": "computertoolset20260801"}, {"type": "texteditor20250728", "name": "strreplacebasededittool"}, {"type": "bash_20250124", "name": "bash"} ], "messages": [ {"role": "user", "content": "Save a picture of a cat to my desktop."} ] }' ` 這類 CLI 內容只是來源文件中的整合範例,不代表可以直接在未檢查的環境中執行。尤其只要同一個 agent loop 另行加入 bash、檔案操作、登入資訊或外部網路,仍需人工核對權限與資料流向。 Browser tool 改用頁面結構 只需要操作網頁時,Anthropic 建議使用 browser use tool,而不是建立完整 desktop environment。Computer use 主要依賴 screenshot 與座標;browser tool 則會讀取頁面結構,讓 Claude 根據 element references 加上 coordinates 鎖定欄位、按鈕或其他 web elements。 這種設計針對的是 pixel-based automation 的常見問題:當網頁排版改變時,原本指向按鈕的座標可能失效。browser tool 仍可處理視覺資訊,但會多取得頁面結構,因此比單純依賴 pixels 更能適應 layout shift。瀏覽器工具的說明見 Browser use tool。 在同一個 tools array 中,Computer use 與 browser use tool 可以並用。兩個 toolset 各自擁有 coordinate frame;若兩者都有 screenshot 或 key 等同名 member,則要用 toolsetname 區分。只要任一 toolset 尚未實作,就必須對相應 block 回傳 iserror: true,不能靜默忽略。 Agent loop 與批次處理 Computer use 的執行核心是 agent loop: 將 Computer tool 與需要 desktop interaction 的 prompt 傳給 Claude。 Claude 回傳 stopreason: "tooluse",其中可能包含一個或多個 tool_use blocks。 應用程式依照 blocks 出現的順序,在自有 container 或 virtual machine 執行每個 member call。 為每個 call 建立對應的 toolresult,保留正確的 tooluseid,並在結果中加入 "toolsetname": "computer"。 screenshot 與 zoom 必須回傳 image;其他 action 可回傳 OK 等短文字。 將所有結果放入新的 user message,再送回 Claude。 Claude 持續請求 actions,直到不再要求 tools,最後改以文字回覆,或達到最大迭代次數。 官方示範的 loop 如下: `python for in range(maxiterations): response = client.messages.create( model=model, max_tokens=4096, messages=messages, tools=TOOLS, ) messages.append({"role": "assistant", "content": response.content}) toolresults = processtool_calls(response) if not tool_results: return messages messages.append({"role": "user", "content": tool_results}) return messages ` 各語言的等價實作包含 Python、TypeScript、C#、Go、Java、PHP 與 Ruby,預設 max_iterations/maxIterations 為 10。這個上限是防止意外 infinite loops 與不可預期 API costs 的 safeguard;loop 會在 Claude 不再要求…

Episode metadata supplied by the publisher feed · Published Aug 21, 2026

Embed this episode

Ready to play

Anthropic 讓 Claude Platform 全面提供 Computer use、browser tool、Skills API 與 Files API

0:00 10:49

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

No similar episodes found.

No similar podcasts found.

Frequently Asked Questions

How long is this episode of EasyVibeCoding Podcast?

This episode is 10 minutes long.

When was this EasyVibeCoding Podcast episode published?

This episode was published on August 21, 2026.

Can I download this EasyVibeCoding Podcast episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.
URL copied to clipboard!