[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-claude-code-dynamic-workflow-ai-harness-zh":3,"article-related-claude-code-dynamic-workflow-ai-harness-zh":31,"series-ai-agent-ef96a410-24bd-4e35-8536-439f21f820e6":83},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":23,"views":27,"created_at":28,"published_at":29,"topic_cluster_id":30},"ef96a410-24bd-4e35-8536-439f21f820e6","claude-code-dynamic-workflow-ai-harness-zh","Claude Code 動態工作流：AI 自寫 Harness","\u003Cp data-speakable=\"summary\">\u003Ca href=\"\u002Ftag\u002Fanthropic\">Anthropic\u003C\u002Fa> 正在讓 \u003Ca href=\"\u002Ftag\u002Fclaude-code\">Claude Code\u003C\u002Fa> 參與寫自己的調度流程，目標是用動態 Harness 取代固定工作流。\u003C\u002Fp>\u003Cp>Anthropic 正把 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002F\" target=\"_blank\" rel=\"noopener\">Anthropic\u003C\u002Fa> 的 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002Fclaude-code\" target=\"_blank\" rel=\"noopener\">Claude Code\u003C\u002Fa> 往更像「會自己編排任務」的方向推進。這次討論的核心，不是模型又多會聊天，而是它開始參與決定任務怎麼拆、工具怎麼叫、步驟怎麼跑。\u003C\u002Fp>\u003Cp>這種做法背後的背景很直接：單一上下文窗口同時塞下規劃、執行、工具回傳，常會讓流程變亂。Anthropic 想解的，是把寫死的控制流程，改成可由模型動態生成的工作流。\u003C\u002Fp>\u003Ch2>發生了什麼\u003C\u002Fh2>\u003Cp>作者把 Anthropic 近來的一串定制流程放在一起看，包括 Research、\u003Ca href=\"\u002Fnews\u002Fopenai-latest-moves-pricing-safety-scale-zh\">安全\u003C\u002Fa>審計、\u003Ca href=\"\u002Ftag\u002Fagent\">Agent\u003C\u002Fa> Teams、\u003Ca href=\"\u002Ftag\u002Fcode-review\">Code Review\u003C\u002Fa>。這些系統都不是一般聊天介面，而是外面先包一層控制器，讓模型先規劃，再按步驟執行。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781035378200-qkm9.png\" alt=\"Claude Code 動態工作流：AI 自寫 Harness\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>原因很現實：如果規劃和執行全塞在同一輪對話裡，資訊很快就會膨脹，工具調用也更容易失控。對複雜任務來說，固定提示詞常常不夠，固定流程也不一定適合每種情境。\u003C\u002Fp>\u003Cp>現在 \u003Ca href=\"\u002Ftag\u002Fclaude\">Claude\u003C\u002Fa> Code 被拿來討論成另一種路線：模型不只產出答案，還能先寫出自己的 Harness，再依那套流程跑任務。這代表工作流不再完全由人預先寫死，而是由模型根據任務型態動態決定。\u003C\u002Fp>\u003Cul>\u003Cli>Research、審計、Agent Teams、Code Review 都屬於定制 Harness。\u003C\u002Fli>\u003Cli>這些流程的共同點，是把規劃和執行拆開。\u003C\u002Fli>\u003Cli>問題核心不是模型不會做事，而是上下文窗口不夠同時裝下全部決策。\u003C\u002Fli>\u003Cli>動態工作流想讓模型參與調度，而不是只產出文字。\u003C\u002Fli>\u003C\u002Ful>\u003Cp>和傳統 agent 架構相比，差別在控制權。以前是工程師先設計好每一步，模型只在框架內回答；現在則是讓模型先判斷任務結構，再選擇合適的執行路徑。\u003C\u002Fp>\u003Cp>這也解釋了為什麼 Anthropic 一直在強調特定任務的流程化設計。當任務越複雜，單靠一個大提示詞硬撐，通常只會換來更長的上下文，而不是更穩定的結果。\u003C\u002Fp>\u003Cp>如果這條路走通，Claude Code 的角色就會更像任務編排器，而不只是寫程式助手。它要處理的不是「回答得好不好」，而是「流程設計得對不對」。\u003C\u002Fp>\u003Ch2>為什麼重要\u003C\u002Fh2>\u003Cp>對開發者來說，這會改變 agent 的設計重心。以前大家多半在比 prompt 寫法，接下來更可能在比控制層、狀態管理、工具路由和失敗回復機制。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781035381141-zbjr.png\" alt=\"Claude Code 動態工作流：AI 自寫 Harness\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>如果模型能根據任務生成合適的工作流，很多場景就不必手工維護一套固定流程。這對內部工具、程式碼審查、資料檢索、研究助手這類高變動任務特別有用，因為它們很少只有一條標準路徑。\u003C\u002Fp>\u003Cp>對產業來說，這也是 Anthropic 把 Claude Code 從單點編碼工具，往更通用的編排層推。真正的競爭不只是誰的模型更會寫，而是誰能更穩定地決定先\u003Ca href=\"\u002Fnews\u002F500-ai-agent-projects-show-where-agents-work-now-zh\">做什麼\u003C\u002Fa>、後做什麼、何時呼叫什麼工具。\u003C\u002Fp>\u003Cp>另一個差異在維護成本。固定 Harness 好處是可控，但每加一種場景就要補一套流程；動態 Harness 若能穩定運作，理論上可以提高復用率，也讓 agent 更容易適配不同任務。\u003C\u002Fp>\u003Cp>問題也很清楚：當 AI 連工作流都能自己寫，下一層競爭就\u003Ca href=\"\u002Fnews\u002Fchocolatey-go-package-policy-installs-zh\">變成\u003C\u002Fa>誰來定規則、誰來審流程、誰來保證它不把步驟跑歪。\u003C\u002Fp>\u003Cp>這不是把控制交出去那麼簡單，而是把「怎麼做事」也變成模型能力的一部分。真正要看的，不是 Claude Code 會不會更像人，而是它能不能穩定把複雜任務拆成可執行的路線。\u003C\u002Fp>","Anthropic 讓 Claude Code 參與寫自己的調度流程，想用動態 Harness 取代固定工作流，重點不在聊天，而在任務編排。","zhuanlan.zhihu.com","https:\u002F\u002Fzhuanlan.zhihu.com\u002Fp\u002F2045777882824325082",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781035378200-qkm9.png","ai-agent","zh","5efa67dd-b9f7-4a2f-8c68-3a4bc6a6b7d9",[17,18,19,20,21,22],"Claude Code","Anthropic","Harness","agent workflow","工具調度","動態工作流",[24,25,26],"Anthropic 正把 Claude Code 從聊天工具推向任務編排系統。","動態 Harness 的重點，是讓模型參與規劃與調度，而不是只輸出答案。","對開發者來說，agent 競爭重心會往控制層、狀態管理與工具路由移動。",1,"2026-06-09T20:02:21.942031+00:00","2026-06-09T20:02:21.93+00:00","4703dc49-ff82-4272-bdfd-a05ff5140912",{"tags":32,"relatedLang":42,"relatedPosts":46},[33,34,36,38,40],{"name":21,"slug":21},{"name":35,"slug":35},"harness",{"name":17,"slug":37},"claude-code",{"name":18,"slug":39},"anthropic",{"name":20,"slug":41},"agent-workflow",{"id":15,"slug":43,"title":44,"language":45},"claude-code-dynamic-workflow-ai-harness-en","Claude Code 动态工作流：AI 自写 Harness","en",[47,53,59,65,71,77],{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":13},"9fb91fbe-64cd-4732-aba7-5b20daacf962","agent-orchestration-enterprise-ai-layer-zh","企業 AI 缺的是編排層","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780984981291-rodj.png","2026-06-09T06:02:30.929215+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":13},"2e389faa-a4ab-4f7a-b6da-c2ba69d5f14b","ai-agents-use-blockchain-trust-layer-zh","AI 代理用區塊鏈當信任層","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780980509390-6s0i.png","2026-06-09T04:48:01.259033+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":13},"1c433948-634b-47e4-a119-dd567203a712","8-rag-patterns-demos-into-prod-zh","8 種 RAG 模式把 Demo 變上線","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780971552397-h12o.png","2026-06-09T02:18:36.130013+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":13},"7d860405-aca6-486b-8de0-1c5193a3b06d","fine-tuning-beats-rag-style-not-facts-zh","當目標是文風不是事實時，微調比 RAG 更有效","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780924689232-5elu.png","2026-06-08T13:17:25.235242+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":13},"3d1e5ef7-8f31-4e57-b286-306825d7f38e","openclaw-small-business-ai-staff-zh","OpenClaw把AI變成夜班員工","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780904888882-6w0v.png","2026-06-08T07:47:27.229503+00:00",{"id":78,"slug":79,"title":80,"cover_image":81,"image_url":81,"created_at":82,"category":13},"0cd44c8d-6ba8-4e6c-851b-d040a5c1a9bd","litellm-rust-minimal-ai-gateway-zh","LiteLLM 推出 Rust 版輕量網關","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1780899485895-xavr.png","2026-06-08T06:17:32.954118+00:00",[84,89,94,99,104,109,114,119,124,129],{"id":85,"slug":86,"title":87,"created_at":88},"4ae1e197-1d3d-4233-8733-eafe9cb6438b","claude-now-uses-your-pc-to-finish-tasks-zh","Claude 開始幫你操作電腦","2026-03-26T07:20:48.457387+00:00",{"id":90,"slug":91,"title":92,"created_at":93},"5bede67f-e21c-413d-9ab8-54a3c3d26227","googles-2026-ai-agent-report-decoded-zh","Google 2026 AI Agent 報告解讀","2026-03-26T11:15:22.651956+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"2987d097-563f-46c7-b76f-b558d8ef7c2b","kimi-k25-review-stronger-still-not-legend-zh","Kimi K2.5 評測：更強，但還不是神作","2026-03-27T07:15:55.277513+00:00",{"id":100,"slug":101,"title":102,"created_at":103},"95c9053b-e3f4-4cb5-aace-5c54f4c9e044","claude-code-controls-mac-desktop-zh","Claude Code 也能操控 Mac 了","2026-03-28T03:01:58.58121+00:00",{"id":105,"slug":106,"title":107,"created_at":108},"dc58e153-e3a8-4c06-9b96-1aa64eabbf5f","cloudflare-100x-faster-ai-agent-sandbox-zh","Cloudflare 的 AI 沙箱跑超快","2026-03-28T03:09:44.142236+00:00",{"id":110,"slug":111,"title":112,"created_at":113},"1c8afc56-253f-47a2-979f-1065ff072f2a","openai-backs-isara-agent-swarm-bet-zh","OpenAI 挺 Isara 的 agent swarm …","2026-03-28T03:15:27.513155+00:00",{"id":115,"slug":116,"title":117,"created_at":118},"7379b422-576e-45df-ad5a-d57a0d9dd467","openai-plan-automated-ai-researcher-zh","OpenAI 想做自動化 AI 研究員","2026-03-28T03:17:42.090548+00:00",{"id":120,"slug":121,"title":122,"created_at":123},"48c9889e-86df-450b-a356-e4a4b7c83c5b","harness-engineering-ai-agent-reliability-2026-zh","駕馭工程：從「馬具」到「作業系統」，AI Agent 可靠性的終極密碼","2026-03-31T06:42:53.556721+00:00",{"id":125,"slug":126,"title":127,"created_at":128},"96d8e8c8-1edd-475d-9145-b1e7a1b02b65","mcp-explained-from-prompts-to-production-zh","MCP 怎麼把提示詞變工作流","2026-04-01T09:24:39.321274+00:00",{"id":130,"slug":131,"title":132,"created_at":133},"f2ca7720-b471-4ce5-9336-2a9ac2a876fd","amazon-bedrock-agents-multi-agent-workflows-zh","Amazon Bedrock Agents 進入多代理工作流","2026-04-01T09:30:29.945429+00:00"]