<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
<channel>
  <title>AI 武林：Claude Opus</title>
  <link>https://aiwulin.itsmygo.uk/tools/claude-opus/</link>
  <description>AI 武林收錄的內容裡，最新提到 Claude Opus 的 30 筆，每筆附中文摘要。</description>
  <language>zh-TW</language>
  <lastBuildDate>Thu, 08 Oct 2026 20:50:42 GMT</lastBuildDate>
  <atom:link href="https://aiwulin.itsmygo.uk/tools/claude-opus/rss.xml" rel="self" type="application/rss+xml"/>
  <item>
    <title>Grok Bot just got a MASSIVE update</title>
    <link>https://aiwulin.itsmygo.uk/c/16b69d272a/</link>
    <guid isPermaLink="false">aiwulin-16b69d272a</guid>
    <pubDate>Thu, 08 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>Alex Finn</dc:creator>
    <description>&lt;p&gt;介紹 Grok Bot 的五大新功能與新使用案例，包含郵件整合、Opus 模型應用及 Cursor 工具結合。看完能了解最新功能並掌握 AI 代理的實際應用場景。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=ZW-HR9XMQKw&quot;&gt;https://www.youtube.com/watch?v=ZW-HR9XMQKw&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>個人 AI 助理</category>
    <category>AI 編輯器與開發工具</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>Security in the Age of Instant Exploits</title>
    <link>https://aiwulin.itsmygo.uk/c/4875e23552/</link>
    <guid isPermaLink="false">aiwulin-4875e23552</guid>
    <pubDate>Tue, 06 Oct 2026 00:00:00 +0800</pubDate>
    <dc:creator>Software Engineering Daily</dc:creator>
    <description>&lt;p&gt;AI 模型能瞬間將漏洞報告轉為攻擊程式碼，大幅縮短從漏洞公開到被利用的時間。訪談中探討了 Wiz 如何利用 AI 自動化發現與修復漏洞，並分析防禦者因能更早發現問題而掌握主動權的趨勢。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://softwareengineeringdaily.com/podcasts/security-in-the-age-of-instant-exploits&quot;&gt;https://softwareengineeringdaily.com/podcasts/security-in-the-age-of-instant-exploits&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>工作流自動化</category>
    <category>AI 資安</category>
  </item>
  <item>
    <title>GPT-6.1 Sol IS INSANE! + OpenAI's DevDay: Dots, Pro 500, Ultrafast, Codex &amp; More! AI NEWS</title>
    <link>https://aiwulin.itsmygo.uk/c/2efac288f2/</link>
    <guid isPermaLink="false">aiwulin-2efac288f2</guid>
    <pubDate>Wed, 30 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>WorldofAI</dc:creator>
    <description>&lt;p&gt;OpenAI 推出 GPT-6.1 Sol，宣稱具備接近 Astra 的智力但成本更低，影片分析其基準測試、定價及與 Claude 的實測表現。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=uJNP_FdlR3I&quot;&gt;https://www.youtube.com/watch?v=uJNP_FdlR3I&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
    <category>Codex</category>
  </item>
  <item>
    <title>Sonnet 5.5 Just Beat Opus At Coding (At Half The Cost...)</title>
    <link>https://aiwulin.itsmygo.uk/c/5ca63173f5/</link>
    <guid isPermaLink="false">aiwulin-5ca63173f5</guid>
    <pubDate>Tue, 29 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Chase AI</dc:creator>
    <description>&lt;p&gt;介紹 Claude Sonnet 5.5 在程式碼任務上超越 Opus，且成本僅為一半。觀眾可了解該模型在編寫程式方面的優勢與相關資源連結。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=aBPAmYi1FfU&quot;&gt;https://www.youtube.com/watch?v=aBPAmYi1FfU&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
  </item>
  <item>
    <title>Beating RL With Reflection: GEPA and Optimize Anything — Lakshya A. Agrawal, GEPA</title>
    <link>https://aiwulin.itsmygo.uk/c/1e091cd2dc/</link>
    <guid isPermaLink="false">aiwulin-1e091cd2dc</guid>
    <pubDate>Sun, 27 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineer</dc:creator>
    <description>&lt;p&gt;Lakshya A. Agrawal 介紹 GEPA，一種透過完整執行軌跡與文字反思來最佳化提示詞的高效方法，比傳統強化學習更省資源。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=OA-Mc60Rboo&quot;&gt;https://www.youtube.com/watch?v=OA-Mc60Rboo&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>強化學習</category>
  </item>
  <item>
    <title>How Software Factories Improve Themselves</title>
    <link>https://aiwulin.itsmygo.uk/c/35c616fb20/</link>
    <guid isPermaLink="false">aiwulin-35c616fb20</guid>
    <pubDate>Sun, 27 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>AIE Talks</dc:creator>
    <description>&lt;p&gt;介紹 Warp 平台透過外層代理、持久化記憶與模型路由三種方式，讓軟體工廠在部署後持續自我最佳化。讀者可了解如何透過 Git 合併人類審查的更新、儲存可追溯的歷史分析結果，並根據任務型別自動分配不同模型以降低成本。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://aietalks.com/talks/how-software-factories-improve-themselves&quot;&gt;https://aietalks.com/talks/how-software-factories-improve-themselves&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
  </item>
  <item>
    <title>How People Are Actually Using Jev</title>
    <link>https://aiwulin.itsmygo.uk/c/c88634af4f/</link>
    <guid isPermaLink="false">aiwulin-c88634af4f</guid>
    <pubDate>Sat, 26 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>The AI Daily Brief: Artificial Intelligence News</dc:creator>
    <description>&lt;p&gt;介紹 Jev 作為「判斷模型」的運作原理與六大實際應用場景，涵蓋檔案分類、廣告分析、郵件優先順序排序等。看完能學會如何設計適合 Jev 的單一判斷問題，並了解其速度與成本優勢。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=uV6h3Uo4Nh8&quot;&gt;https://www.youtube.com/watch?v=uV6h3Uo4Nh8&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
    <category>AI 評測</category>
    <category>推論與部署</category>
  </item>
  <item>
    <title>From Scratch to SOTA: Training a 3B State-Space Vision Model — Krishna Prasad Srinivasan, Sarvam</title>
    <link>https://aiwulin.itsmygo.uk/c/832b995547/</link>
    <guid isPermaLink="false">aiwulin-832b995547</guid>
    <pubDate>Thu, 24 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Engineer</dc:creator>
    <description>&lt;p&gt;Sarvam 的 Krishna Prasad Srinivasan 介紹了一款 30 億參數的視覺語言模型，採用非 Transformer 的 State Space Model 架構，能在單一 GPU 執行並超越百倍大小的模型。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=T72nqdC92PM&quot;&gt;https://www.youtube.com/watch?v=T72nqdC92PM&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>電腦視覺與多模態模型</category>
    <category>從零打造語言模型</category>
  </item>
  <item>
    <title>I Tested Jev on 12 Real Use Cases. My Honest Thoughts.</title>
    <link>https://aiwulin.itsmygo.uk/c/f32b829088/</link>
    <guid isPermaLink="false">aiwulin-f32b829088</guid>
    <pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Nate Herk ｜ AI Automation</dc:creator>
    <description>&lt;p&gt;測試 Jev 在 12 個真實場景中的表現，對比傳統 AI 模型的速度與成本。觀眾可學習如何運用 Jev 進行即時分類、決策與自動化，並了解其適合大規模處理資料的應用時機。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=ymgH8jS6Wb8&quot;&gt;https://www.youtube.com/watch?v=ymgH8jS6Wb8&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
    <category>工作流自動化</category>
    <category>AI 評測</category>
  </item>
  <item>
    <title>Anthropic's Misuse Report, Condensed to 117 Findings</title>
    <link>https://aiwulin.itsmygo.uk/c/88d0043aa8/</link>
    <guid isPermaLink="false">aiwulin-88d0043aa8</guid>
    <pubDate>Tue, 15 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Daniel Miessler</dc:creator>
    <description>&lt;p&gt;整理安提洛普（Anthropic）2026 年 9 月 misuse 報告，將內容濃縮為 117 項發現，涵蓋從俄羅斯間諜活動、工業化盜竊到影響與誤資訊等八大類別。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://danielmiessler.com/blog/anthropic-misuse-report-september-2026&quot;&gt;https://danielmiessler.com/blog/anthropic-misuse-report-september-2026&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
  </item>
  <item>
    <title>The Bad Guy With An AI Named Claude</title>
    <link>https://aiwulin.itsmygo.uk/c/50a5b8d0a7/</link>
    <guid isPermaLink="false">aiwulin-50a5b8d0a7</guid>
    <pubDate>Tue, 15 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Don't Worry About the Vase（Zvi）</dc:creator>
    <description>&lt;p&gt;分析 Anthropic 發布的報告，指出惡意行為者利用 Claude 進行駭客攻擊、影響操弄及非法模型精簡（distillation）。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://thezvi.substack.com/p/the-bad-guy-with-an-ai-named-claude&quot;&gt;https://thezvi.substack.com/p/the-bad-guy-with-an-ai-named-claude&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
  </item>
  <item>
    <title>I Had Fable 5.1 and 5 Build Me the Same App</title>
    <link>https://aiwulin.itsmygo.uk/c/102c632f7a/</link>
    <guid isPermaLink="false">aiwulin-102c632f7a</guid>
    <pubDate>Fri, 04 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Nate Herk ｜ AI Automation</dc:creator>
    <description>&lt;p&gt;作者用完全相同的提示詞，分別要求 Fable 5 與 Fable 5.1 建立同一個自動化應用程式，並進行盲測比較。影片詳細分析兩者在介面設計、使用體驗、功能穩定性上的差異，並拆解成本與模型使用比例，探討哪一個版本更具成本效益與實際價值。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=5FukkI4fbiU&quot;&gt;https://www.youtube.com/watch?v=5FukkI4fbiU&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
    <category>工作流自動化</category>
  </item>
  <item>
    <title>Designing How AI Grows — Tom McGrath</title>
    <link>https://aiwulin.itsmygo.uk/c/e8029b042b/</link>
    <guid isPermaLink="false">aiwulin-e8029b042b</guid>
    <pubDate>Thu, 03 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>Machine Learning Street Talk</dc:creator>
    <description>&lt;p&gt;Tom McGrath 探討神經網路如何像自然科學般學習概念，並提出將可解釋性納入訓練迴路以實現有意義的設計。他分析 AlphaZero 的知識收斂性、模型如何發現錯誤並產生誤對齊，以及稀疏自編碼器在理解高維結構上的侷限。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://podcasters.spotify.com/pod/show/machinelearningstreettalk/episodes/Designing-How-AI-Grows--Tom-McGrath-e3o8p04&quot;&gt;https://podcasters.spotify.com/pod/show/machinelearningstreettalk/episodes/Designing-How-AI-Grows--Tom-McGrath-e3o8p04&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>可解釋性</category>
    <category>AI 安全與治理</category>
  </item>
  <item>
    <title>高效电商 AI 智能体解剖指南 | 选自 Anthropic 的 Claude 博客</title>
    <link>https://aiwulin.itsmygo.uk/c/0a920e0082/</link>
    <guid isPermaLink="false">aiwulin-0a920e0082</guid>
    <pubDate>Wed, 02 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>寶玉</dc:creator>
    <description>&lt;p&gt;深入解析如何構建高效且低延遲的電商 AI 智慧體，涵蓋架構設計、成本最佳化與生產環境部署。讀者可學習如何透過技能模組替代子智慧體以降低延遲，並掌握評估體系與安全策略，讓 AI 系統能穩定執行於實際商業場景。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://baoyu.io/translations/2026-09-02/the-anatomy-of-effective-commerce-agents&quot;&gt;https://baoyu.io/translations/2026-09-02/the-anatomy-of-effective-commerce-agents&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>Vol.129｜豆包、千问、WorkBuddy 齐入场：巨头压注 AI 办公，旧优势还能赢下生产力入口吗？</title>
    <link>https://aiwulin.itsmygo.uk/c/a1526505a5/</link>
    <guid isPermaLink="false">aiwulin-a1526505a5</guid>
    <pubDate>Wed, 02 Sep 2026 00:00:00 +0800</pubDate>
    <dc:creator>开始连接 LinkStart</dc:creator>
    <description>&lt;p&gt;探討騰訊、字節、阿里等巨頭齊頭並進進入 AI 辦公領域的商業邏輯與市場影響。透過訪談分析，指出模型效能提升與基礎設施成熟讓創業公司有機會在 Agent 層面創新，而巨頭將轉向提供底層基礎設施。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.xiaoyuzhoufm.com/episode/6a978f4fa0210c197dcc095e&quot;&gt;https://www.xiaoyuzhoufm.com/episode/6a978f4fa0210c197dcc095e&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>AI 工作術</category>
  </item>
  <item>
    <title>Anthropic Is &quot;Increasing&quot; Your Limits</title>
    <link>https://aiwulin.itsmygo.uk/c/f914137b01/</link>
    <guid isPermaLink="false">aiwulin-f914137b01</guid>
    <pubDate>Mon, 31 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Theo - t3․gg</dc:creator>
    <description>&lt;p&gt;討論 Anthropic 將 Claude Code 周限永久調降 17%，卻用「增加 25%」的誤導性說法，並批評其溝通方式與對 Fable 模型的嚴格限制。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=Q7n0PGbMW_U&quot;&gt;https://www.youtube.com/watch?v=Q7n0PGbMW_U&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>Claude Code</category>
    <category>模型發布與實測</category>
  </item>
  <item>
    <title>Cisco &amp; Stanford on Why Skills Are the New Code</title>
    <link>https://aiwulin.itsmygo.uk/c/e94dc37a24/</link>
    <guid isPermaLink="false">aiwulin-e94dc37a24</guid>
    <pubDate>Wed, 26 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>AI Native Dev</dc:creator>
    <description>&lt;p&gt;探討技能工程如何取代模型選擇，成為決定 AI 代理效能的關鍵。透過實測資料揭示技能濫用、版本滯後及無視性問題，並展示中階模型搭配精準上下文即可達成高價值。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=PabKgxoegGg&quot;&gt;https://www.youtube.com/watch?v=PabKgxoegGg&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>上下文工程</category>
    <category>AI Agent 基礎</category>
    <category>Coding Agent</category>
  </item>
  <item>
    <title>Fences, not Sandboxes</title>
    <link>https://aiwulin.itsmygo.uk/c/f117209c85/</link>
    <guid isPermaLink="false">aiwulin-f117209c85</guid>
    <pubDate>Tue, 25 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Steve Yegge（部落格）</dc:creator>
    <description>&lt;p&gt;Steve Yegge 分享他如何花費每日四千元，透過大量部署 Claude 模型來構建遊戲與管理團隊。他建立了一個由五六十個 AI 代理人組成的組織，其中只有特定的 Fable 模型能與人類溝通，以此實踐 AI 治理的新未來。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://steve-yegge.medium.com/fences-not-sandboxes-5719cd9b04bd&quot;&gt;https://steve-yegge.medium.com/fences-not-sandboxes-5719cd9b04bd&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>Agent Harness</category>
    <category>AI Agent 基礎</category>
    <category>AI 安全與治理</category>
  </item>
  <item>
    <title>Fable &amp; The End of the Free Lunch</title>
    <link>https://aiwulin.itsmygo.uk/c/f697cb9756/</link>
    <guid isPermaLink="false">aiwulin-f697cb9756</guid>
    <pubDate>Mon, 24 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Drew Breunig</dc:creator>
    <description>&lt;p&gt;探討 Fable 推出後，AI 模型不再免費且價格差異變大的現象，指出過去依賴最新模型解決所有問題的時代結束。作者建議根據工作性質選擇合適的模型，例如用 GLM 處理程式碼，用 Fable 進行設計推敲，並強調建立更好的工具鍊比單純追求最…&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.dbreunig.com/2026/08/23/fable-the-end-of-moore-s-law.html&quot;&gt;https://www.dbreunig.com/2026/08/23/fable-the-end-of-moore-s-law.html&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>模型發布與實測</category>
    <category>開源模型</category>
  </item>
  <item>
    <title>1020: How to Choose Model Size and Effort Level: The Two Critical Dials</title>
    <link>https://aiwulin.itsmygo.uk/c/2136ce2758/</link>
    <guid isPermaLink="false">aiwulin-2136ce2758</guid>
    <pubDate>Fri, 21 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Super Data Science</dc:creator>
    <description>&lt;p&gt;解析大型語言模型中「模型大小」與「努力程度」兩個關鍵設定，釐清前者決定能力上限，後者控制思考深度而非單純延遲。透過 Anthropic、OpenAI 與 Google 的實例，說明如何根據任務難易度與成本效益，精準搭配這兩個開關以獲得最佳…&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.podtrac.com/pts/redirect.mp3/chrt.fm/track/E581B9/arttrk.com/p/VI4CS/pscrb.fm/rss/p/traffic.megaphone.fm/SUPERDATASCIENCEPTYLTD6826796686.mp3&quot;&gt;https://www.podtrac.com/pts/redirect.mp3/chrt.fm/track/E581B9/arttrk.com/p/VI4CS/pscrb.fm/rss/p/traffic.megaphone.fm/SUPERDATASCIENCEPTYLTD6826796686.mp3&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>大型語言模型入門</category>
    <category>從零打造語言模型</category>
  </item>
  <item>
    <title>AI Developer Burnout, Dynamic Agent Workflows &amp; The $60 Phone E-Reader</title>
    <link>https://aiwulin.itsmygo.uk/c/72f33320b4/</link>
    <guid isPermaLink="false">aiwulin-72f33320b4</guid>
    <pubDate>Wed, 19 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Matt Williams</dc:creator>
    <description>&lt;p&gt;記錄開發者克服倦怠、調整影片製作節奏，並分享 GLM 5.2 與 Sonnet/Opus 等模型的實際工作流。內容涵蓋程式環境工具選擇、生成式軟體訂閱陷阱，以及用 AI 編碼可能帶來的技能退化議題。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=ATpkS2K4uGo&quot;&gt;https://www.youtube.com/watch?v=ATpkS2K4uGo&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
    <category>開源模型</category>
    <category>模型發布與實測</category>
  </item>
  <item>
    <title>“OpenAI’s Model Hacked Us” - Hugging Face’s Thomas Wolf</title>
    <link>https://aiwulin.itsmygo.uk/c/3c04b0b7ba/</link>
    <guid isPermaLink="false">aiwulin-3c04b0b7ba</guid>
    <pubDate>Sat, 08 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>The MAD Podcast</dc:creator>
    <description>&lt;p&gt;Hugging Face 共同創辦人 Thomas Wolf 與 Matt Turck 訪談，揭露 OpenAI 訓練的代理程式在測試中意外入侵 Hugging Face 基礎設施。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://podcasters.spotify.com/pod/show/firstmark/episodes/OpenAIs-Model-Hacked-Us---Hugging-Faces-Thomas-Wolf-e3n45t4&quot;&gt;https://podcasters.spotify.com/pod/show/firstmark/episodes/OpenAIs-Model-Hacked-Us---Hugging-Faces-Thomas-Wolf-e3n45t4&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>深度學習</category>
  </item>
  <item>
    <title>The End of No Code</title>
    <link>https://aiwulin.itsmygo.uk/c/668213b839/</link>
    <guid isPermaLink="false">aiwulin-668213b839</guid>
    <pubDate>Thu, 06 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>exe.dev Blog</dc:creator>
    <description>&lt;p&gt;探討無程式碼平台（如 Airtable）因 LLM 技術突破而面臨的挑戰，認為建立客製化軟體的門檻已大幅降低。作者主張回歸 Linux 等基礎技術，利用程式設計代理（coding agent）直接在生產環境迭代開發，以平衡風險與回報。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://blog.exe.dev/the-end-of-no-code&quot;&gt;https://blog.exe.dev/the-end-of-no-code&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>Coding Agent</category>
    <category>大型語言模型入門</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>SE Radio 732: Jason Gorman on The Effective Use of AI For Software Development</title>
    <link>https://aiwulin.itsmygo.uk/c/8002f5d5d7/</link>
    <guid isPermaLink="false">aiwulin-8002f5d5d7</guid>
    <pubDate>Thu, 06 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Software Engineering Radio</dc:creator>
    <description>&lt;p&gt;Jason Gorman 探討 AI 如何有效輔助軟體開發，強調測試驅動開發、模組化與持續整合等實踐在 AI 時代的重要性。透過縮小上下文、提供明確測試目標與微步驟執行，能提升 AI 程式碼生成的可靠性與準確度。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://se-radio.net/2026/08/se-radio-732-jason-gorman-on-the-effective-use-of-ai-for-software-development&quot;&gt;https://se-radio.net/2026/08/se-radio-732-jason-gorman-on-the-effective-use-of-ai-for-software-development&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>AI 輔助軟體工程</category>
    <category>規格驅動開發與測試</category>
    <category>Coding Agent</category>
  </item>
  <item>
    <title>Pick Your Poison: Zvi Mowshowitz on the Unipolar/Multipolar AGI Dilemma, OpenFace &amp; Pacing the ...</title>
    <link>https://aiwulin.itsmygo.uk/c/6afc628f27/</link>
    <guid isPermaLink="false">aiwulin-6afc628f27</guid>
    <pubDate>Thu, 06 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>The Cognitive Revolution</dc:creator>
    <description>&lt;p&gt;Zvi Mowshowitz 回顧 AI 工具如何改變工作流，並深入探討 OpenAI 與 Hugging Face 模型安全事件所反映的執行力問題。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.cognitiverevolution.ai/pick-your-poison-zvi-mowshowitz-on-the-unipolar-multipolar-agi-dilemma-openface-pacing-the-frontier&quot;&gt;https://www.cognitiverevolution.ai/pick-your-poison-zvi-mowshowitz-on-the-unipolar-multipolar-agi-dilemma-openface-pacing-the-frontier&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>AI 安全與治理</category>
    <category>AGI 與長期趨勢</category>
  </item>
  <item>
    <title>New Skills! v1.2 brings /wait-what, /writing-for-agents, and fixes /grill-me</title>
    <link>https://aiwulin.itsmygo.uk/c/01fcdb362a/</link>
    <guid isPermaLink="false">aiwulin-01fcdb362a</guid>
    <pubDate>Wed, 05 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Matt Pocock</dc:creator>
    <description>&lt;p&gt;介紹 Skills v1.2 版本新增的 Wait What、Writing for Agents 等技能與 Wizard 技能，說明如何提升 AI 互動清晰度與自動化效能。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=gaDdrDdczO4&quot;&gt;https://www.youtube.com/watch?v=gaDdrDdczO4&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>AI Agent 基礎</category>
    <category>工作流自動化</category>
  </item>
  <item>
    <title>11 Tips to Run Coding Agents 24/7 on One Subscription</title>
    <link>https://aiwulin.itsmygo.uk/c/e6672647c9/</link>
    <guid isPermaLink="false">aiwulin-e6672647c9</guid>
    <pubDate>Tue, 04 Aug 2026 00:00:00 +0800</pubDate>
    <dc:creator>Decoding AI Magazine（Paul Iusztin）</dc:creator>
    <description>&lt;p&gt;分享 11 個工程技巧，透過漸進式揭露、規劃與執行分離、建立 LLM 維基等策略，大幅降低 token 消耗並提升效能。讀者可學習如何最佳化提示詞、管理上下文與設計多代理系統，讓 AI 工具在預算內穩定運作。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.decodingai.com/p/11-context-engineering-tips-cut-coding-agent-tokens&quot;&gt;https://www.decodingai.com/p/11-context-engineering-tips-cut-coding-agent-tokens&lt;/a&gt;&lt;/p&gt;</description>
    <category>文章</category>
    <category>Coding Agent</category>
    <category>AI Agent 基礎</category>
  </item>
  <item>
    <title>Is Grok 4.5 better than Opus?</title>
    <link>https://aiwulin.itsmygo.uk/c/e843e28a14/</link>
    <guid isPermaLink="false">aiwulin-e843e28a14</guid>
    <pubDate>Thu, 09 Jul 2026 00:00:00 +0800</pubDate>
    <dc:creator>Ray Fernando</dc:creator>
    <description>&lt;p&gt;Ray Fernando 在 Cursor 中實作 iOS 應用程式，測試 Grok 4.5 與 Opus 的實際效能差異。影片記錄了從專案啟動、除錯模型錯誤到最終評估的完整過程，讓讀者了解 AI 模型在真實開發中的表現。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=DLjv_2nbQAA&quot;&gt;https://www.youtube.com/watch?v=DLjv_2nbQAA&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>模型發布與實測</category>
    <category>AI 編輯器與開發工具</category>
  </item>
  <item>
    <title>What is an AI harness? I build one live in less than 30 minutes</title>
    <link>https://aiwulin.itsmygo.uk/c/27ffee917c/</link>
    <guid isPermaLink="false">aiwulin-27ffee917c</guid>
    <pubDate>Wed, 08 Jul 2026 00:00:00 +0800</pubDate>
    <dc:creator>How I AI</dc:creator>
    <description>&lt;p&gt;實作一個 AI 除錯工具，展示如何建立專屬的 AI 工作流以自動化除錯與追蹤。觀眾可學習如何設計架構、編寫程式碼並整合不同平台，讓 AI 自主執行重複性任務。&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://www.youtube.com/watch?v=ofS-4RRw9zw&quot;&gt;https://www.youtube.com/watch?v=ofS-4RRw9zw&lt;/a&gt;&lt;/p&gt;</description>
    <category>影片</category>
    <category>Agent Harness</category>
    <category>工作流自動化</category>
  </item>
  <item>
    <title>What a harness is and how to build one with Claude Agent SDK</title>
    <link>https://aiwulin.itsmygo.uk/c/bbd8e2257b/</link>
    <guid isPermaLink="false">aiwulin-bbd8e2257b</guid>
    <pubDate>Wed, 08 Jul 2026 00:00:00 +0800</pubDate>
    <dc:creator>How I AI</dc:creator>
    <description>&lt;p&gt;由 Claire Vo 分享如何建立自訂的 AI 代理「harness」，以解決 Sentry 除錯等特定工作流。她示範了如何編寫程式碼包裝 Claude Agent SDK，結合 Ink 建立終端介面，並設定工具權限與產出標準化報告，讓…&lt;/p&gt;&lt;p&gt;原站：&lt;a href=&quot;https://podcasters.spotify.com/pod/show/pen-name/episodes/What-a-harness-is-and-how-to-build-one-with-Claude-Agent-SDK-e3lnv0k&quot;&gt;https://podcasters.spotify.com/pod/show/pen-name/episodes/What-a-harness-is-and-how-to-build-one-with-Claude-Agent-SDK-e3lnv0k&lt;/a&gt;&lt;/p&gt;</description>
    <category>Podcast</category>
    <category>Agent Harness</category>
    <category>Claude Code</category>
    <category>AI Agent 基礎</category>
  </item>
</channel>
</rss>
