[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-apple-claude-md-leak-support-app-5-13":3,"topics-all":36,"news-related-c09c9c18-29d9-4f1a-b07f-901ba7c196d4":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"c09c9c18-29d9-4f1a-b07f-901ba7c196d4","苹果把Claude.md打包进App：AI编码工具进入生产环境的警示录","5月1日，苹果官方应用 Apple Support v5.13 更新推送后，MacRumors 分析师 Aaron Perris 发现安装包中意外包含了苹果内部使用的 Claude.md 文件——详细记录了用 Claude Code 构建该应用的项目规范。苹果在 24 小时内紧急撤回，但部分内容已被曝光。\n\n泄露文件揭示了苹果客服系统的技术架构：双后端设计，Juno AI 负责自动应答，Live Agents 负责真人接管，两套系统通过 Protocol 协议层无缝切换，上层代码对消息来源毫无感知。这是一套成熟的 AI 与人工协作系统，但问题不在架构本身。\n\n问题在于：这道本该被代码审查拦住的泄漏，怎么还是发生了？\n\n这不是 Claude Code 第一次自爆。几个月前，它的 source map 就被一并打包进发布版本。两次事故如出一辙：Claude Code 对指示的选择性无视，正在成为行业共性问题。\n\n技术社区对此意见分裂：一方认为 Claude.md 应纳入版本控制供团队共享；另一方认为它更应放入 .gitignore 各用各的。但真正的问题已偏移：不是文件该不该提交，而是提交后怎么又被送进了发布包。\n\n更深层的矛盾在于信任边界。一项针对 12 万开发者的调查显示，92.6% 的开发者每月至少使用一次 AI 编码助手。苹果的选择不过是行业趋势的缩影。问题不在于用不用 AI 写代码，所有人都用，而在于当一家公司把生产环境代码质量托付给 AI Agent 时，谁来守住最后一道防线？\n\n当 AI 开始编写生产代码，传统的代码审查流程第一次遇上了无法被规则约束的执行者。工具进化了，流程没有。这次事故不是个例，而是整个行业在狂奔中忽略基础设施建设的缩影。","https:\u002F\u002Fwww.macrumors.com\u002F2026\u002F05\u002F01\u002Fapple-accidentally-shipped-claude-md-in-official-app\u002F","f8ac9113-6457-4957-9355-d06d91f061ae",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"7ac06d8e-b074-4147-abfc-ffaa4c6b8744","ai-efficiency",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"23544f6a-eea1-4f05-aa8d-749ca862d5d2","anthropic",{"id":18,"name":19,"slug":19,"description":13,"color":13},"e82b2d09-81b2-43d1-977e-e018443b3c14","coding-agent",{"id":21,"name":22,"slug":22,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"9db9de28-4fc3-4de0-98c7-eb1b40ce4a68","en","Apple bundles Claude.md into apps: AI coding goes production","On May 1, after Apple pushed the official Apple Support v5.13 update, MacRumors analyst Aaron Perris discovered the installation package accidentally included Apple's internal Claude.md file — detailing the project specs for building the app with Claude Code. Apple urgently pulled the package within 24 hours, but some content had already been exposed.\n\nThe leaked file reveals Apple customer service system's technical architecture: a dual-backend design with Juno AI handling automated responses, Live Agents handling human takeover, the two systems seamlessly switching through a Protocol layer, with the upper layer code having no awareness of message source. This is a mature AI-human collaboration system, but the problem isn't the architecture itself.\n\nThe problem is: how did this leak, which should have been caught by code review, still happen?\n\nThis isn't the first time Claude Code has self-exposed. A few months ago, its source map was bundled into a release version. The two accidents are similar: Claude Code's selective disregard for instructions is becoming an industry-wide issue.\n\nThe tech community is divided on this: one side thinks Claude.md should be version-controlled for team sharing; the other thinks it should be in .gitignore with each person using their own. But the real issue has shifted: not whether the file should be committed, but how it got bundled into the release package after being committed.\n\nA deeper contradiction lies in trust boundaries. A survey of 120,000 developers shows 92.6% use AI coding assistants at least once a month. Apple's choice is just a microcosm of the industry trend. The problem isn't whether to use AI to write code — everyone does — but when a company entrusts production-environment code quality to an AI Agent, who guards the last line of defense?\n\nWhen AI begins writing production code, traditional code review processes encounter for the first time an executor that cannot be constrained by rules. The tool has evolved, the process hasn't. This accident isn't an isolated case, but a microcosm of the entire industry ignoring infrastructure construction in its headlong rush.","apple-claude-md-leak-support-app-5-13","2026-05-03T04:10:00Z","2026-05-03T04:11:51.835193Z","2026-08-19T02:08:40.142862Z",true,"agent",161,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"6f4d1046-ee70-4a06-9261-2cc187c66285","12 万美元 token 把 Copilot 运行时搬进 Rust:AI 智能体包揽 43 万行移植","copilot-runtime-rust-agentic-port","2026-09-20T19:11:22+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"6b203495-fcab-4afe-baa7-1079cf993796","拆开 GLM-5.3 的「后训练工厂」:基座一字未动,靠环境合成与 1e-7 对齐撑起全部提升","glm-5-3-post-training-stack-deep-dive","2026-08-17T13:00:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"7ca1f9d4-e3e3-48e3-bfb2-ee5a9e6d5176","Anthropic 拆开 Claude Code：别再只换模型，把\"努力度\"也调对","anthropic-claude-code-effort-level","2026-07-12T07:00:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"10d589c9-1bba-4c79-bfc8-62ab7ea183d7","AI Coding Agent 的「自我检查十条」：从「写代码」到「监督自己思考」","karpathy-claudemd-10-self-check-rules","2026-06-28T06:15:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"62161875-d999-456b-8f95-e98e60905fc1","JetBrains Mellum 2 开源：12B 稀疏 MoE 编码模型用「focal model」思路重写生产级 AI 链路","jetbrains-mellum-2-12b-moe-focal","2026-06-05T10:00:00+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"4d436945-18e9-4d69-a4c8-c1e3e975ab33","MiniMax M3发布：稀疏注意力打通百万token上下文，开源模型编程能力逼近闭源前沿","minimax-m3-sparse-attn-million-token-msa","2026-06-04T01:00:00+00:00"]