2026-07-10（内容获取于 07/10 05:16）

Patch for Windows Defender 0-day could allow attackers to fill hard disk

Ars Technica · 07/10 04:52

The feud between NightmareEclipse and Microsoft shows no signs of resolving soon.

中文介绍 微软Windows Defender的零日漏洞补丁被指存在缺陷，可能允许攻击者耗尽用户硬盘空间。安全研究员NightmareEclipse与微软之间的争议尚未解决。

The ChatGPT browser is already dead

The Verge · 07/10 04:34

OpenAI is already shutting down ChatGPT Atlas, its browser that could do tasks for you on your behalf, less than a year after launching it. Atlas was announced in October, but as part of its wave of news about ChatGPT Work today, the company confirmed that it will be "sunsetting" Atlas and is target

中文介绍 OpenAI宣布将关闭其去年10月推出的AI浏览器ChatGPT Atlas。该浏览器旨在为用户执行任务，但运营不足一年便被“逐步淘汰”。此举是该公司围绕ChatGPT Work系列新闻的一部分。

The floatable, powerful Soundcore Boom 2 speaker is over half off

The Verge · 07/10 04:32

Anker’s Soundcore Boom 2 is fully waterproof. Bluetooth speakers with big sound and great features are hard to find for under $100, with most offerings being some variation of the same basic (and often small) design. Thankfully, through July 10th Woot has the Anker Soundcore Boom 2 on sale for $69.9

中文介绍 Anker旗下的Soundcore Boom 2蓝牙音箱正在Woot上进行促销，折扣超过一半，截止日期为7月10日。该音箱具有完全防水、可漂浮和强劲音质等特点，以低于100美元的价格出售，性价比突出。

Allstate accuses Broadcom of auditing it because it quit VMware, CA

Ars Technica · 07/10 04:28

Broadcom accuses Allstate of dodging VMware audits.

中文介绍 Allstate指控博通(Broadcom)因其停止使用VMware和CA产品而对其进行审计。博通则反驳称Allstate一直在回避VMware的审计。双方在企业IT采购和合规问题上存在争议。

Anthropic found a hidden space where Claude puzzles over concepts

MIT Tech Review AI · 07/10 04:22

The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what’s really going on inside large language models as they answer questions or carry out tasks. What they found ranges from the mundane to the unnerving. Researchers at the company built a tool called the

中文介绍 人工智能公司Anthropic开发了一种新技术，使其能更清晰地观察大型语言模型（LLM）Claude在回答问题或执行任务时内部的运作机制。研究人员发现了一个“隐藏空间”，其中模型处理概念的方式既有平凡之处，也有令人不安的层面。

Humanoid robots controlled by surgeons did world-first operation on live pigs

Ars Technica · 07/10 04:12

Preclinical trial is testing the feasibility of humanoid robots in surgery.

中文介绍 在一项世界首创的术前试验中，由外科医生控制的类人机器人在活猪身上成功实施了手术。该试验旨在评估类人机器人在外科手术中的可行性，标志着医疗机器人技术的重要进展。

Google will now tell you if an ad was made with AI

The Verge · 07/10 04:11

You can see if ads on Google Search, Google Discover, and YouTube were made or edited using AI from a new section in Google's "My Ad Center," as reported earlier by TechCrunch. The update, announced on Thursday, adds a "created or edited with AI" label under the "how this ad was made" tab. Users can

中文介绍 谷歌宣布，其广告平台将新增AI内容识别标签。用户可在Google搜索、Google Discover和YouTube上，通过“我的广告中心”的新部分，查看广告是否由AI生成或编辑，标签为「created or edited with AI」。此举旨在提高广告透明度。

Meta enters the crowded AI coding battle with Muse Spark 1.1

TechCrunch · 07/10 03:40

Meta's pitch to users is Spark's ability to handle large agentic workloads, fix bugs, and help with large code migrations — the kind of automation that enterprises are increasingly turning to AI companies to provide.

Charles Hudson shares the common mistakes he’s seen after investing in 500+ startups

TechCrunch · 07/10 03:26

In this week’s episode of Build Mode, Isabelle Johannessen talks with Precursor Ventures' Charles Hudson about the headwinds facing early-stage founders today and the most common mistakes founders should avoid in order to get funded.

中文介绍 Precursor Ventures的Charles Hudson在Build Mode访谈中分享了他投资超过500家初创公司后，观察到的早期创始人常犯的错误。他讨论了当前早期创始人面临的挑战，并提供了获得融资应避免的常见失误。

Judge doesn't like Elon Musk settlement with SEC, but says court can't block it

Ars Technica · 07/10 03:10

Judge reluctantly approves $1.5M settlement with SEC over Twitter stock violation.

中文介绍 法官不情愿地批准了埃隆·马斯克(Elon Musk)与美国证券交易委员会(SEC)之间达成的150万美元和解协议，涉及推特股票违规行为。法官表示不认同和解内容，但法院无法阻止。

New York Times says OpenAI hid evidence in ChatGPT copyright trial

TechCrunch · 07/10 03:05

News publishers say OpenAI hid tools and datasets that could identify copyrighted journalism in ChatGPT outputs, escalating their lawsuit with a new motion for sanctions.

中文介绍 《纽约时报》指控OpenAI在ChatGPT版权诉讼中隐瞒证据，包括可识别ChatGPT输出中受版权保护新闻内容的工具和数据集。新闻出版商因此提出新的制裁动议，使法律纠纷升级。

Slate Auto teams up with Crayola to color its EV truck

TechCrunch · 07/10 03:01

Slate has an answer for owners who have always want to drive a truck with bright crayon colors.

OpenAI may have made a fatal misstep in copyright fight with news orgs

Ars Technica · 07/10 02:57

OpenAI may be sanctioned for hiding, deleting ChatGPT logs in NYT copyright fight.

中文介绍 在与《纽约时报》的版权纠纷中，OpenAI被指控伪造无法搜索训练数据的能力，并隐藏了数十亿条ChatGPT日志。此举可能导致OpenAI面临制裁，被认为是其在法律战中的严重失误。

FanDuel sent a video from star athlete Bryce Harper to a customer with a gambling problem

TechCrunch · 07/10 02:50

The video call highlights the murky relationship between professional athletes and gambling apps.

中文介绍 FanDuel被曝光向一位存在赌博问题的用户发送了明星运动员布莱斯·哈珀(Bryce Harper)的视频。此事件凸显了职业运动员与赌博应用程序之间复杂且可能存在问题的关系，引发对行业营销道德的关注。

Mercor is in talks for a $20B valuation

TechCrunch · 07/10 02:42

A new $20 billion valuation would be a giant step up from the $10 billion valuation it reached in October.

中文介绍 Mercor目前正就获得200亿美元估值进行谈判。若此次估值达成，将是该公司在去年10月达到100亿美元估值之后的一次巨大飞跃，显示出其快速增长的势头。

MadsLorentzen/ai-job-search

TypeScript · ★ 18,613 · 🍴 5,402 · 📈 3,728 stars today

AI-powered job application framework built on Claude Code. Fork it, fill in your profile, and let Claude evaluate jobs, tailor CVs, write cover letters, and prepare you for interviews.

中文介绍 `MadsLorentzen/ai-job-search` 是一个基于 Claude AI 构建的智能求职框架。用户只需填写个人档案，该系统就能自动评估职位匹配度、量身定制简历、撰写求职信，甚至为面试提供准备。它旨在利用 AI 显著提升求职效率和质量，帮助求职者优化申请流程，提高成功几率。

SmartlyDressedGames/U3-SDK

C# · ★ 1,967 · 🍴 249 · 📈 541 stars today

Source code for Unturned, a free open-world zombie survival sandbox game.

中文介绍 该项目提供了免费开放世界僵尸生存沙盒游戏 Unturned 的完整源代码。它不仅是一个供玩家深入了解游戏机制的宝库，更面向游戏开发者和模组制作者。利用此SDK，用户可以学习游戏开发技术、创建自定义内容（mod），甚至进行二次开发，极大地扩展了游戏的可能性和社区创造力。

addyosmani/agent-skills

JavaScript · ★ 75,761 · 🍴 8,148 · 📈 2,582 stars today

Production-grade engineering skills for AI coding agents.

中文介绍 为 AI 编码代理提供生产级的工程技能，旨在提升 AI 代理在软件开发中的实际能力和可靠性。它可能包含代码生成、测试、重构等实用工具和方法，帮助开发者构建更高效、更可靠的智能编程助手。适用于 AI 工程师和软件开发团队。

VoltAgent/awesome-design-md

★ 99,519 · 🍴 11,567 · 📈 1,233 stars today

A collection of DESIGN.md files analysis by popular brand design systems. Drop one into your project and let coding agents generate a matching UI.

中文介绍 此项目收集并分析了来自知名品牌设计系统的 DESIGN.md 文件集合，旨在为AI编码代理提供输入。通过这些结构化的设计规范，编码代理能够自动生成匹配的 UI 界面，显著加速开发流程。它特别适用于需要快速将设计系统转换为代码、自动化前端开发，以及探索AI辅助UI生成的开发者和设计师。

iOfficeAI/OfficeCLI

C# · ★ 13,311 · 🍴 904 · 📈 1,923 stars today

OfficeCLI is the first and best Office suite purpose-built for AI agents to read, edit, and automate Word, Excel, and PowerPoint files. Free, open-source, single binary, no Office installation required.

中文介绍 `iOfficeAI/OfficeCLI` 是一款专为 AI 代理设计的开源 Office 套件，它允许 AI 代理无缝读取、编辑和自动化处理 Word、Excel 和 PowerPoint 文件。该工具以单一二进制文件形式提供，无需安装传统 Office 软件，极大地简化了 AI Agent 与 Office 文档的交互流程。它为开发者构建能够自动处理文档的智能系统提供了高效且免费的解决方案。

wonderwhy-er/DesktopCommanderMCP

TypeScript · ★ 6,558 · 🍴 779 · 📈 185 stars today

This is MCP server for Claude that gives it terminal control, file system search and diff file editing capabilities

中文介绍 DesktopCommanderMCP 是一个为 Claude AI 模型设计的 MCP（Master Control Program）服务器，赋予其对本地桌面的控制能力。通过该服务器，Claude 能够执行终端命令、搜索文件系统及进行基于差异的文本编辑。它解决了大型语言模型与真实操作系统交互能力有限的问题，使得 AI Agent 不仅能理解和生成文本，还能实际操作计算机完成复杂任务。开发者可利用此工具为 Claude 等 LLM 构建具备高级自动化和系统交互能力的桌面级 AI 助理。

anthropics/claude-cookbooks

Jupyter Notebook · ★ 47,092 · 🍴 5,554 · 📈 194 stars today

A collection of notebooks/recipes showcasing some fun and effective ways of using Claude.

中文介绍 这是 Anthropic 官方出品的 Claude 大型语言模型使用指南和示例代码集合。项目以 Jupyter Notebook 形式，展示了多种有趣且高效的 Claude 应用场景和 Prompt 工程技巧。旨在帮助开发者、研究人员或对大型语言模型感兴趣的用户，更好地理解和利用 Claude 的能力，加速基于 Claude 的应用开发和实验。

vxcontrol/pentagi

Go · ★ 19,316 · 🍴 2,617 · 📈 543 stars today

Fully autonomous AI Agents system capable of performing complex penetration testing tasks

中文介绍 Pentagi 是一个完全自主的 AI Agents 系统，专注于执行复杂的渗透测试任务。它利用人工智能的能力，自动化地发现系统漏洞、评估安全风险，并模拟攻击路径，从而提高测试效率和准确性。适用于网络安全专家、渗透测试人员和企业安全团队，旨在实现更智能、更高效的安全防御体系。

unclecode/crawl4ai

Python · ★ 71,769 · 🍴 7,375 · 📈 195 stars today

🚀🤖 Crawl4AI: Open-source LLM Friendly Web Crawler & Scraper. Don't be shy, join here: https://discord.gg/jP8KfhDhyN

中文介绍 Crawl4AI 是一款开源的“LLM友好”网页爬虫与数据抓取工具，专门为大型语言模型设计。它能够高效地从网页中提取信息，并以LLM易于理解和处理的格式呈现数据，解决AI模型数据输入的痛点。目标用户是AI开发者、数据科学家以及需要为LLM应用（如RAG、内容生成）获取高质量训练或推理数据的团队。

imthenachoman/How-To-Secure-A-Linux-Server

★ 29,031 · 🍴 1,935 · 📈 306 stars today

An evolving how-to guide for securing a Linux server.

中文介绍 这份项目提供了一份持续更新的 Linux 服务器安全加固操作指南。该指南详细列出了从基础设置到高级配置的全面步骤和建议，旨在帮助用户有效提升服务器的防御能力，防范潜在的网络威胁，确保系统运行的安全性。主要面向系统管理员、DevOps工程师、网络安全初学者以及任何需要部署和维护安全 Linux 服务器的个人或团队。

huxingyi/autoremesher

C++ · ★ 2,324 · 🍴 171 · 📈 405 stars today

Automatic quad remeshing tool

中文介绍 autoremesher 是一个自动四边形网格重拓扑工具。在 3D 建模领域，它能将任意拓扑的 3D 模型自动转换为主要由四边形构成的网格，解决手动重拓扑耗时且复杂的问题。四边形网格对于动画、雕刻和细分曲面具有更好的形变特性和更优的视觉效果。该工具极大地简化了 3D 模型的后期处理流程。3D 艺术家、游戏开发者或从事 CAD 数据处理的工程师将从中受益，用于优化模型拓扑、提高渲染和动画质量。

bradautomates/claude-video

Python · ★ 6,621 · 🍴 763 · 📈 727 stars today

Give Claude the ability to watch any video. /watch downloads, extracts frames, transcribes, hands it all to Claude.

中文介绍 该项目赋予 Claude AI 模型观看和理解视频的能力。它通过下载视频、提取关键帧和生成文字转录，将处理后的内容提供给 Claude 进行分析。解决了大型语言模型原生不支持视频输入的问题，使用户能利用 Claude 进行视频摘要、内容问答或深度分析。适用于 AI 开发者和视频内容分析场景。

prisma/prisma

TypeScript · ★ 46,867 · 🍴 2,327 · 📈 373 stars today

Next-generation ORM for Node.js & TypeScript | PostgreSQL, MySQL, MariaDB, SQL Server, SQLite, MongoDB and CockroachDB

中文介绍 Prisma 是一个专为 Node.js 和 TypeScript 设计的新一代 ORM（对象关系映射器）。它通过提供类型安全的查询构建器和直观的 API，简化了开发者与数据库的交互，解决了传统 ORM 的复杂性和类型安全问题。Prisma 支持 PostgreSQL、MySQL、MariaDB、SQL Server、SQLite、MongoDB 和 CockroachDB 等主流数据库，极大提升了全栈开发效率。适用于构建高性能、类型安全且易于维护的 Node.js/TypeScript 后端应用及 API 服务。

kyutai-labs/pocket-tts

Python · ★ 6,930 · 🍴 697 · 📈 273 stars today

A TTS that fits in your CPU (and pocket)

中文介绍 `kyutai-labs/pocket-tts` 是一个轻量级的文本转语音（TTS）系统，其核心优势在于能够在 CPU 上高效运行，并强调其“口袋化”的极低资源占用特性。它旨在为需要本地、离线且对计算资源有严格限制的应用场景提供解决方案，例如嵌入式设备、移动应用或边缘计算项目。该项目提供了一个兼顾性能与便携性的 TTS 引擎。

asgeirtj/system_prompts_leaks

JavaScript · ★ 55,059 · 🍴 8,990 · 📈 1,135 stars today

Extracted system prompts from Anthropic - Claude Fable 5, Opus 4.8, Claude Code, Claude Design. OpenAI - ChatGPT 5.5 Thinking, GPT 5.5 Instant, Codex. Google - Gemini 3.5 Flash, 3.1 Pro, Antigravity. xAI - Grok, Cursor, Copilot, VS Code, Perplexity, and more. Updated regularly.

中文介绍 system_prompts_leaks 汇集了 Anthropic (如 Claude Fable 5, Opus 4.8)、OpenAI (如 ChatGPT 5.5 Thinking, GPT 5.5 Instant) 和 Google (如 Gemini 3.5 Flash, 3.1 Pro) 等主流大语言模型被“泄露”的系统 Prompt。此项目为研究者、开发者和安全专家提供了深入了解这些模型内部指令和行为模式的宝贵资源，有助于进行 prompt 工程优化和安全分析。

The Task Economy - Data will be the next $1 Trillion Category

@EverettRandle · 19.4K 粉丝 · 484.8K 阅 · 507 赞 · 56 转

07/08 00:15

The Token Economy When we talk about AI today, tokens are king. Specifically, inference tokens have emerged as the primary proxy for tracking the growth of the AI ecosystem. Public companies report

中文介绍 EverettRandle 提出「任务经济」概念，并预测未来数据将成为价值万亿美元的市场新类别。他认为，随着AI应用日益普及，经济价值将从传统的Token消耗转向完成具体任务的能力及数据本身的价值创造与积累，预示着AI驱动的经济转型。

Own the Outer Loop

@addyosmani · 405.4K 粉丝 · 396.4K 阅 · 547 赞 · 69 转

07/09 02:44

In the past year, the conversation around agentic engineering has moved to harnesses and loops, fleets and software factories. My 2c is engineers need to own the outer loop - the accountability for

中文介绍 博主 addyosmani 提出了「拥有外循环」（Own the Outer Loop）的核心概念。该概念可能旨在强调在某种工程或系统设计背景下，个人或团队需要对更宏观、更广阔的循环或流程承担责任和掌控力，而非局限于内部操作。具体语境未在推文开头明确，但指向一种工程管理或责任范式。

Continual Learning for Agents

@pirroh · 15.0K 粉丝 · 314.8K 阅 · 540 赞 · 59 转

07/06 21:10

Everyone talks about Continual Learning as if it means one thing only: updating model weights. But there's an inconvenient truth about the agent ecosystem — the vast majority of agents in production

中文介绍 博主pirroh分享了关于“Agent持续学习”（Continual Learning for Agents）的讨论。内容可能涉及如何在Agent系统中实现持续学习、面临的挑战或最佳实践。具体细节需查看原文，但预示了对AI Agent领域核心学习机制的探讨。

Everyone is wrong about open source AI in the enterprise

@thejessezhang · 59.4K 粉丝 · 274.2K 阅 · 512 赞 · 54 转

07/06 23:31

The prevailing story right now is that open source is eating the enterprise. The capability gap between the best closed and open models has shrunk to low single digits. A third of the Fortune 500 has

中文介绍 博主thejessezhang对“企业级开源AI”的普遍认知提出挑战，认为多数人对此存在误解。内容可能深入分析开源AI在企业应用中的实际情况、潜在的盲点或被忽视的复杂性，旨在提供一个更全面和批判性的视角来理解这一趋势。

Getting started with loops

@ClaudeDevs · 564.6K 粉丝 · 252.4K 阅 · 2.9K 赞 · 239 转

07/07 03:08

There’s a lot of talk right now about "designing loops" instead of prompting your coding agent. If you spend some time on X trying to pin down what a loop actually is, you'll come across multiple

中文介绍 ClaudeDevs 针对 AI 开发中的「循环」概念提供了入门指导。推文可能探讨了如何在 AI 代理设计中应用循环结构，以实现迭代优化或复杂任务处理，旨在帮助开发者理解这一基础工作流。

This prompt will change your life

@EXM7777 · 123.5K 粉丝 · 233.6K 阅 · 554 赞 · 38 转

07/07 22:31

for most people, fable has a few hours left to live... this is the smartest model that ever existed and everyone's spending its last night building one more app i made it do something 100x more

中文介绍 EXM7777 分享了一个其声称「将改变生活」的提示词。尽管推文未透露具体内容，但暗示该提示词具有极强的实用价值和颠覆性效果，能够帮助用户在日常或工作中实现重大突破，引发了社区对高效提示词工程的关注和讨论。

How to become an applied AI engineer

@eyad_khrais · 20.2K 粉丝 · 225.8K 阅 · 508 赞 · 38 转

07/07 23:42

I wrote this as the guide I would have wanted before making the transition into Applied AI Engineering. The role broadly overlaps with traditional software engineering, but it adds a few important

中文介绍 eyad_khrais 撰写了一份「应用AI工程师」转型指南，旨在帮助有志者进入该领域。他指出，应用AI工程师与传统软件工程师角色有重叠，但需掌握更多关键技能，例如如何将AI模型与实际业务场景结合、处理数据管道、优化部署及性能等。这份指南对职业规划和技能树构建提供了宝贵参考。

The Heretic’s Guide to AI’s Stars Part III: AI Demand, Offshore Financing, & Compression Too

@michaeljburry · 2.0M 粉丝 · 209.2K 阅 · 508 赞 · 51 转

07/08 08:23

“They are just flying empty airplanes around.” - Son NVIDIA is the North Star, Orion, the whole Milky Way. NVIDIA is benefitting from strong demand, but is selling into a concentrated set of buyers

中文介绍 michaeljburry 对AI行业发展提出独到见解，指出NVIDIA虽作为「北极星」般的存在，市场需求旺盛，但其销售对象高度集中。他引用“空转飞机”比喻，暗示市场存在虚假繁荣的风险。推文还探讨了AI需求、离岸融资及数据压缩技术在当前市场环境下的作用与挑战。

The Fable Loop Library: 25 Workflows on Autopilot

@EXM7777 · 123.5K 粉丝 · 208.2K 阅 · 525 赞 · 47 转

07/04 23:43

i'm going to teach you how to run Fable 5 on autopilot, using my own library of loops and goals... 25 workflows, each with a prompt and the exact tool it plugs into the method follows karpathy's

中文介绍 博主发布「Fable Loop Library」，提供 25 个 Fable 自动化工作流。该库旨在帮助用户实现 Fable 操作的自动驾驶，有效提升工作效率，简化复杂任务处理。

Career advice in the age of AI

@philhchen · 9.1K 粉丝 · 179.1K 阅 · 516 赞 · 34 转

07/03 05:25

AI models get better at anything you can write a loss function for, and school is mostly loss functions: well-defined problems graded against known answers. Therefore, the valuable work of the next

中文介绍 该帖子围绕AI时代下的职业发展提供建议。博主旨在分享面对人工智能技术进步时，个人应如何规划职业路径、提升自身价值，以适应未来就业市场的变化和挑战。

Expanding Managed Agents in Gemini API: background tasks, remote MCP and more

@GoogleAIStudio · 184.1K 粉丝 · 165.4K 阅 · 504 赞 · 62 转

07/08 00:38

Today we’re announcing new capabilities in Managed Agents in Gemini API, including background execution, remote MCP server integration, custom function calling and refreshing credentials across

中文介绍 GoogleAIStudio 宣布 Gemini API 的 Managed Agents 迎来多项新功能。更新包括支持后台执行，实现远程 MCP 服务器集成，允许自定义函数调用，以及跨项目刷新凭证。这些增强功能旨在提升开发者在使用 Gemini API 构建智能代理时的灵活性和效率，简化复杂工作流的实现。

Your AI, your growth

@arthurmensch · 74.0K 粉丝 · 162.7K 阅 · 539 赞 · 76 转

07/04 05:31

Of course you need to use open-source models if you’re an enterprise leader. Close model providers, that are now forcing data retention, are gaining immense leverage on your business if you don’t. As

中文介绍 该推文仅以“Your AI, your growth”作为预览内容，未提供具体细节或论证，因此无法生成详细摘要。推测博主旨在强调AI对个人或组织成长的重要性，但具体观点、方法或案例均未提及。

You have 1 day to clone Fable 5 into Opus 4.8.

@alex_prompter · 277.4K 粉丝 · 125.9K 阅 · 595 赞 · 77 转

07/07 02:25

Tomorrow is the last day Fable 5 sits inside your plan for free. After July 7 it moves to pay-per-use credits, and most people are about to spend the week arguing over whether it's worth keeping. That

中文介绍 博主alex_prompter发布紧急通知，提醒用户必须在一天内将Fable 5克隆（clone）到Opus 4.8。这可能暗示Fable 5即将面临服务调整或费用变化，建议用户及时采取行动以保留其配置或数据，以免造成不便或损失。

Model and effort in Claude Code: knowing more vs. trying harder

@ClaudeDevs · 564.6K 粉丝 · 95.9K 阅 · 1.5K 赞 · 100 转

07/09 00:55

Claude Code gives you two settings that both seem to "make the answer better": the model, and the effort level. But what do these actually do to the output? And how do you know whether to reach for a

中文介绍 ClaudeDevs 探讨了在利用 Claude Code 进行开发时，是深入理解模型本身「知道更多」还是通过反复尝试和优化工程「更努力」更能提升效率。这一讨论聚焦于提示词工程、模型能力边界与开发者介入程度之间的平衡，旨在为开发者提供关于如何最大化AI编程工具潜力的策略性思考。

Do this on your last day with Fable

@EXM7777 · 123.5K 粉丝 · 86.9K 阅 · 510 赞 · 37 转

07/06 23:48

You can extract Fable 5's intelligence out of the model before it's gone... and i'm going to teach you the 5 workflows that do it, with every prompt ready to paste because tomorrow Fable 5 leaves your

中文介绍 EXM7777 分享了一个关于 Fable 平台的实用建议或操作，特别指出是在使用 Fable 的“最后一天”应该完成的事项。具体内容可能涉及数据处理、账户管理或告别前的特殊功能体验。

Accurate, Interdisciplinary and Transparent Structure-property Understanding with Deep Native Structural Reasoning

👍 71

07/08 08:00

Structure-property relationships are foundational to biology, chemistry and materials science, where function, reactivity and physical response emerge from spatial, chemical and periodic organization. Mechanistically explaining these relationships requires interpreting structural evidence through sc

中文介绍 这篇论文提出一种名为「深度原生结构推理」（Deep Native Structural Reasoning）的方法，旨在准确、跨学科且透明地理解生物学、化学和材料科学中的结构-性质关系。该方法通过解释结构排列来阐明功能、反应性和物理响应如何从空间、化学和周期性组织中产生。

Co-LMLM: Continuous-Query Limited Memory Language Models

👍 0

07/08 08:00

Limited memory language models (LMLMs) externalize factual knowledge during pretraining to a knowledge base (KB), rather than memorizing it in their weights. During generation, the model then fetches knowledge from the KB as needed. This recently introduced paradigm provides multiple advantages, inc

中文介绍 论文提出「Co-LMLM」，即连续查询有限内存语言模型，改进了现有有限内存语言模型（LMLMs）的范式。LMLMs在预训练期间将事实知识外部化到知识库（KB）中，而非存储于模型权重内，并在生成时按需从KB获取知识，从而提升了多方面的能力。

ECGLight: Compute-Light Framework For Paper ECG Digitization and Myocardial Infarction Screening

👍 0

07/08 08:00

Electrocardiography (ECG) is one of the most widely used tests for diagnosing cardiovascular disease. Yet several remote clinics still utilize paper ECG printouts for their analysis due to limited connectivity and computational capacity. As a result, vast numbers of physical ECGs obtained in remote

中文介绍 论文提出「ECGLight」框架，旨在为资源有限的偏远诊所提供计算轻量化的纸质心电图（ECG）数字化及心肌梗死筛查方案。该框架能够有效处理大量纸质ECG记录，以应对这些地区因连接和计算能力限制而无法使用数字设备的问题，改善心血管疾病诊断。

Scaling Mixture-of-Experts Video Pretraining for Embodied Intelligence

👍 39

07/08 08:00

Despite the recent promise in robot control, video generative models suffer from a domain mismatch due to their primary focus on content creation. For example, their design inherently prioritizes visual fidelity and creativity over computational efficiency and physical realism. In this work, we pres

中文介绍 论文指出，现有视频生成模型在机器人控制（具身智能）中存在领域不匹配问题，因其设计侧重视觉逼真度和创造性，而非计算效率和物理真实性。为解决此问题，该研究提出扩展专家混合模型（MoE）进行视频预训练，旨在更好地适应具身智能任务的需求，提升机器人控制能力。

Max Out GRPO Signal: Adaptive Trace Prefix Control for Hard Reasoning Problems

👍 1

07/08 08:00

Group Relative Policy Optimization (GRPO) stalls on a model's hardest problems: when no rollout in a group succeeds, the group-relative advantages vanish and the problem contributes no gradient, wasting the frontier examples we most want to learn from. Prepending a correct prefix of a reference solu

中文介绍 论文指出，在处理模型最困难的问题时，「组相对策略优化」（GRPO）会陷入停滞，因为组内若无成功推演，组相对优势会消失，导致无梯度贡献。为解决此问题，研究提出一种名为「自适应轨迹前缀控制」的方法，通过预置正确前缀来“最大化GRPO信号”，从而有效利用前沿示例进行学习。

Does Bielik Know What It Doesn't Know? Activation Dispersion Separates Entity Familiarity from Factual Reliability Across Model Scale

👍 0

07/08 08:00

Large language models hallucinate most about entities they have never seen. We ask whether a model's activations betray entity familiarity before a single answer token is generated, and whether that signal predicts the factual reliability of the answers. On four Polish Bielik models (1.5B-11B parame

中文介绍 论文探讨大型语言模型（LLMs）是否能在生成答案前，通过激活分散度信号揭示其对实体的熟悉程度，并预测答案的事实可靠性。研究发现，LLMs主要对未曾见过的实体产生幻觉。作者在四种波兰语「Bielik」模型上进行实验，旨在区分实体熟悉度与事实可靠性，以减少模型幻觉。

Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops

👍 0

07/08 08:00

AI systems increasingly participate in their own improvement: revising their outputs, adapting their own harnesses during deployment, training on data they generate, and, increasingly, conducting AI research itself. This literature is described under a vocabulary ("self-refine," "self-reward," "self

中文介绍 论文探讨AI系统中的「递归式自我改进」概念，从有限的自我完善扩展到自主研究循环。AI系统正越来越多地参与自身改进，包括修订输出、在部署中适应自身、利用生成数据进行训练，乃至自主进行AI研究。该研究旨在系统梳理并定义这一不断发展的领域。

Dual Latent Memory in Vision-Language-Action Models for Robotic Manipulation

👍 44

07/08 08:00

Mainstream Vision-Language-Action (VLA) models predict actions primarily from the current observation under a Markovian assumption, thus struggling with long-horizon, temporally dependent tasks. Existing memory-augmented VLAs either expand the observation window or retrieve history from the memory b

中文介绍 论文指出，主流视觉-语言-动作（VLA）模型因马尔可夫假设，难以处理长期、时序依赖的机器人操作任务。为解决此问题，该研究提出一种「双重潜在记忆」（Dual Latent Memory）机制，用于VLA模型。此机制旨在通过更有效的记忆整合，帮助机器人模型克服现有内存增强VLA模型的局限，更好地执行复杂任务。

Infinite Worlds with Versatile Interactions

👍 21

07/08 08:00

We present LingBot-World 2.0 (also known as LingBot-World-Infinity), an advanced iteration of LingBot-World featuring four distinct upgrades. (1) Our model achieves an unbounded interaction horizon while maintaining consistent output quality, benefiting from a carefully crafted causal pretraining pa

中文介绍 论文发布「LingBot-World 2.0」（亦称LingBot-World-Infinity），这是LingBot-World的升级版本，具有四项显著改进。该模型通过精心设计的组件，实现了无界限的交互范围，同时保持了持续一致的输出质量，从而能处理更广泛、更复杂的任务场景，拓展了AI系统的互动潜力。

Single-Rollout Asynchronous Optimization for Agentic Reinforcement Learning

👍 4

07/08 08:00

Reinforcement learning (RL) is becoming increasingly important for post-training large language models (LLMs). Previous RL pipelines for LLMs were mostly synchronous and batch-interleaved, which is inefficient for long-horizon agentic tasks. Recently, asynchronous RL has emerged as a more efficient

中文介绍 强化学习（RL）对大型语言模型（LLMs）的后期训练日益重要。鉴于以往LLMs的RL管线多为同步批处理交错模式，对长期代理任务效率低下，论文提出「单轨迹异步优化」方法，用于代理式强化学习。该方法有望显著提升处理复杂、长周期LLMs任务的效率和性能。

TimEE: End-to-end Time Series Classification via In-Context Learning

👍 10

07/08 08:00

Time series classification (TSC) is dominated by a two-stage paradigm: train a feature encoder -- either from scratch on the target dataset or via pretraining on large corpora -- and then fit a task-specific classifier on top. While effective, this decoupling optimizes representation learning indepe

中文介绍 论文指出，时间序列分类（TSC）目前主要采用两阶段范式，即训练特征编码器后适配分类器。为克服这种解耦优化表示的局限性，该研究提出名为「TimEE」的端到端时间序列分类方法，通过「上下文学习」（In-Context Learning）直接进行分类，旨在提高TSC的效率和性能。

Smooth Operator: A Real-Time Sampling-Based Algorithm for Kinematic Hand Retargeting

👍 0

07/08 08:00

Advances in learning-based robotic manipulation, such as Vision-Language-Action (VLA) models and Video Action Models (VAMs), heavily rely on high-quality teleoperation data. Their capabilities are strictly upper-bounded by the quality of the underlying human demonstrations. Current gradient-based re

中文介绍 论文指出，基于学习的机器人操作（如VLA和VAM模型）高度依赖高质量遥操作数据，其能力受限于人类演示质量。为此，研究提出「Smooth Operator」算法，这是一种实时、基于采样的「运动学手部重定向」方法。该算法旨在优化数据质量，进而提升机器人操纵任务的性能和流畅度。

The Blind Curator: How a Biased Judge Silently Disables Skill Retirement in Self-Evolving Agents

👍 0

07/08 08:00

A self-evolving agent retires its bad skills by watching them fail, so what happens when the judge cannot see the failures? Skill retirement is the structural constraint that keeps a growing library from drifting below the no-skill baseline, but its guarantee assumes an unbiased reward, which is fal

Reason Less, Verify More: Deterministic Gates Recover a Silent Policy-Violation Failure Mode in Tool-Using LLM Agents

👍 0

07/08 08:00

Tool-using LLM agents can violate the very policies they are deployed to enforce while appearing to complete the task successfully. In policy-permissive environments, a tool may execute any well-formed call even when the corresponding state transition is forbidden by domain policy. The result is a s

Sparse Delta Memory: Scaling the State of Linear RNNs through Sparsity

👍 5

07/08 08:00

Linear attention models allow a fixed state size and a fixed amount of compute per token. However, due to their limited state size, linear attention models fall behind in long-context recall compared to softmax-attention-based transformer architectures. Increasing the state size of linear attention

From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM Agents

👍 0

07/08 08:00

Tool utilization enables Large Language Model (LLM) agents to interact with the real world and resolve complex tasks. However, existing agent frameworks predominantly rely on static toolsets composed of granular atomic actions (e.g., basic file I/O or single-turn search), which forces agents to rein

TouchWorld: A Predictive and Reactive Tactile Foundation Model for Dexterous Manipulation

👍 0

07/08 08:00

Dexterous manipulation in everyday environments requires both anticipation and reaction: a robot must predict how contact should evolve while rapidly correcting local errors caused by slip, misalignment, unstable grasping, or force mismatch. Vision and language provide semantic and geometric guidanc

Progressive Crystallization: Turning Agent Exploration into Deterministic, Lower-Cost Workflows in Production

👍 0

07/08 08:00

AI agents deployed for IT operations are typically permanent cost centers because every execution requires full LLM inference, even for previously solved problems. This paper introduces progressive crystallization, a lifecycle that treats agent exploration as a discovery mechanism rather than a perm

Ego-Human Motion Prediction with 3D-Aware LLM

👍 0

07/08 08:00

Anticipating human motion from an egocentric perspective is fundamental for proactive assistance in AR/VR, human-robot collaboration, and embodied AI. While recent works incorporate language as a semantic prior to reduce the ill-posed nature of egocentric forecasting, they largely neglect the 3D spa

WAM-TTT: Steering World-Action Models by Watching Human Play at Test Time

👍 0

07/08 08:00

Steering robot foundation models (RFMs) toward new task variants or user-preferred behaviors remains challenging, often requiring additional robot demonstrations, task-specific fine-tuning, or long-context conditioning. We present WAM-TTT, a test-time training framework for steering world action mod

Weak-to-Strong Generalization via Direct On-Policy Distillation

👍 0

07/08 08:00

Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training. As models scale, post-training itself becomes a bottleneck.

RWGBench: Evaluating Scholarly Positioning in Related Work Generation

👍 0

07/08 08:00

Large language models have shown strong fluency in scientific writing, yet the evaluation of related work generation (RWG) remains limited. Existing RWG evaluations largely inherit summarization-oriented metrics, using lexical or semantic similarity to reference sections as proxies for quality. Howe

The Signs Were Always There: Training-Free Concept Detection and Steering in Raw Transformer Dimensions

👍 2

07/08 08:00

The standard basis of transformer hidden states is a training-free, architecture-general feature basis for detecting concepts and, in language models, steering them; with no learned dictionary. Individual dimensions act as binary registers read one at a time: their signs (+/-1) encode content, their

MARGIN: Margin-Aware Regularized Geometry for Imbalanced Vulnerability Detection

👍 0

07/08 08:00

Software vulnerability detection is critical for ensuring software security and reliability. Despite recent advances in deep learning, real-world vulnerability datasets suffer from two severe challenges: frequency imbalance and difficulty imbalance. We reinterpret these challenges from an embedding

WildCity: A Real-World City-Scale Testbed for Rendering, Simulation, and Spatial Intelligence

👍 4

07/07 08:00

Humans can navigate an unfamiliar city and gradually form a coherent spatial mental map spanning tens of square kilometers. Can AI build spatial representations at a comparable scale? Although recent foundation models have advanced scene reconstruction and embodied intelligence, scaling to entire ci

Trees from Marginals: Autoregressive drafting with factorized priors

👍 0

07/07 08:00

Speculative decoding greatly increases the interactivity of autoregressive language models by trading off computation for extra tokens generated in a single forward pass. Factorized draft models are especially efficient because they predict future-token marginals in parallel, but their independence

Vision Language Action (VLA) Models for Unmanned Aerial Robotics and Bimanual Manipulation: A Review

👍 0

07/07 08:00

Vision Language Action (VLA) models unify visual perception, natural-language understanding, and action generation within a single foundation model, allowing a robot to follow instructions such as fold the towel or fly to the red building directly from camera images. Because VLAs inherit world knowl

RoboSnap: One-Shot Real-to-Sim Scene Generation for Generalizable Robot Learning and Evaluation

👍 10

07/07 08:00

Recovering real-world scenes as interactive simulation environments can enable generalizable robot learning and reproducible policy evaluation. However, constructing scenes that are both physically stable and visually faithful remains slow and expensive. In this work, we present RoboSnap, a real-to-

NativeMEM: Native Memory Compression for Long-Horizon Robotic Manipulation

👍 0

07/07 08:00

How can pretrained Vision-Language-Action (VLA) models retain long-horizon visual histories with high-frequency updates without sacrificing efficiency? Existing approaches rely on external memory management, which restrains either the memory horizon or the reactiveness of pretrained policies. To thi

Pelican-VLA 0.5: Attending Before Acting Benefits Generalization

👍 0

07/07 08:00

In this report, we present Pelican-VLA 0.5, a unified VLA model that integrates vision-language understanding, future-frame generation, and action prediction within a single architecture. Pelican-VLA 0.5 achieves attention-level generalization: without object annotations, segmentation masks, attenti

Opper AI

07/10 05:09

The european AI gateway for agents

中文介绍 Opper AI 定位为欧洲人工智能代理的门户平台。它旨在为用户提供连接和管理AI代理的服务，并作为欧洲AI代理生态系统的关键入口。

GPT-Live

07/10 05:09

Full-duplex voice for ChatGPT

中文介绍 GPT-Live 为 ChatGPT 提供了全双工语音功能。这意味着用户可以与 ChatGPT 进行实时、双向的语音对话，无需等待对方说完，大幅提升交互体验。

Timbal AI

07/10 05:09

Build AI agents, workflows, and apps in one stack

中文介绍 Timbal AI 提供一体化开发堆栈，使用户能够构建AI代理、工作流和应用程序。它旨在简化AI产品的开发过程，将多种功能整合在单一平台中，提高开发效率。

Perfai Security

07/10 05:08

Find & fix live vulnerabilities in Vibe Apps with 1-prompt.

中文介绍 Perfai Security 是一款针对 Vibe 应用的安全工具，它能通过「一键提示」的方式，快速发现并修复在运行中的应用漏洞。该产品旨在简化安全漏洞的管理和解决流程，提升应用安全性。

Lispr

07/10 05:06

Hold a key, speak, and Lispr writes it anywhere

中文介绍 Lispr 是一款语音输入工具，用户只需按住一个按键并说话，Lispr 就能将语音内容实时转换为文字并输入到任何应用程序中。它旨在提高打字效率和输入便捷性。

ARKAD Wallet

07/10 05:05

Control your budget with voice

中文介绍 ARKAD Wallet 是一款支持语音控制的预算管理工具。用户可以通过语音指令来管理个人预算、记录收支、查看账单，实现便捷直观的财务控制，简化记账过程。

Toyo

07/10 05:04

Exec assistant who lives in iMessage and calls your phone

中文介绍 Toyo 是一款智能行政助理，它常驻于 iMessage 应用中，并能通过拨打电话来协助用户处理事务。这款AI助理旨在为用户提供便捷高效的执行支持，处理日常行政工作。

Monogram AI

07/10 05:02

AI with a visual and interactive interface

中文介绍 Monogram AI 提供具有可视化和交互界面的AI服务。它旨在通过直观友好的图形界面，让用户能够更轻松地与AI进行交互和操作，降低AI使用门槛，提升用户体验。

LemonLime

07/10 05:02

Automates your existing workflows with a single prompt.

中文介绍 LemonLime 是一款自动化工具，用户可以通过一个简单的提示词来自动化现有工作流程。它旨在利用AI技术，极大地简化和加速日常任务的执行，提高工作效率。

Auriko

07/10 04:58

Trading desk for LLM calls

中文介绍 Auriko 提供一个大型语言模型（LLM）调用的「交易台」。它旨在帮助用户管理、优化并有效利用对各种LLM的API调用，可能涉及成本控制、性能优化及资源分配。

Grok 4.5 + Cursor Just Shocked Everyone

07/10 01:51

中文介绍 根据YouTube视频标题，xAI的AI模型Grok已更新至4.5版本，并与代码编辑器Cursor共同出现，暗示可能带来令人关注的新进展或功能整合。视频标题强调此举令「所有人震惊」，但具体技术细节或发布内容在摘要中未提供。

Claude Fable 5 Just Changed Design Forever

07/07 01:37

I Created the Ultimate Jarvis AI Assistant (It’s So Easy)

07/01 09:40

Poweful Agents inside Slack

06/25 06:39

AI Agents Just Changed Forever: GLM 5.2, Codex Skills, Claude & Cursor

06/21 21:30

9 AI Agent Skills To Get Ahead of 99% of People

06/18 10:27

How to build high quality internal tools with Retool

06/18 05:21

SpaceX Just Bought Cursor for $60B. It’s About to Take OVER.

06/17 05:31

Why the Government Just Killed Claude Fable 5

06/13 11:50

This is Why They Banned Claude Fable 5

06/13 02:28

GPT-5.6 还有 5 分钟上线

07/10 00:56

19 回复 · 程序员 节点

终于轮到我了， Codex 又又又重置了

07/09 18:06

5 回复 · 程序员 节点

grok4.5 太强了

07/09 16:56

5 回复 · 程序员 节点

codex 的额度是不是出问题了，变来变去的

07/09 16:44

6 回复 · 程序员 节点

clash verge 能不能导入分流规则的配置文件？

07/09 04:59

6 回复 · Apple 节点

Shadowrocket 支持 Tailscale 了，那支持 split dns 吗？

07/09 03:16

9 回复 · Apple 节点

Apple 美区礼品卡购买问题

07/08 14:15

30 回复 · Apple 节点

2026 年 07 月购买了 Quantumult X

07/08 13:46

72 回复 · Apple 节点

iOS 有什么办法能屏蔽懂球帝的广告

07/08 13:39

15 回复 · Apple 节点

卖还是不卖？

07/08 09:53

36 回复 · Apple 节点

How GitHub gave every repository a durable owner

07/10 04:47

5 points · 0 comments

Show HN: RandoFont, a browser for Google Fonts

07/10 04:46

Inevitably when starting a new web project I want to pick the perfect font. This is just a browser that shows Google Fonts, randomly, and lets me tag favorites.I built this years ago and just used AI to give it a minor facelift.

GLM 5.2 is nearly as accurate as a human book keeper

07/10 02:29

132 points · 79 comments

How to Start a Ruby Meetup

07/10 02:27

38 points · 10 comments

GPT-5.6

07/10 01:04

https://deploymentsafety.openai.com/gpt-5-6/gpt-5-6.pdfhttps://developers.openai.com/api/docs/guides/latest-modelhttps://x.com/levie/status/2075287443411222628, https://xcancel.com/levie/status/207528

ChatGPT Work

07/10 01:03

281 points · 125 comments

Wildcard (YC W25) Is Hiring a Founding Engineer

07/10 01:00

1 points · 0 comments

AI content is everywhere on social media, especially LinkedIn

07/09 23:50

127 points · 111 comments

Launch HN: Context.dev (YC S26) – API to get structured data from any website

07/09 23:28

Hi Hacker News, I’m Yahia. I built Context.dev (https://www.context.dev/) to make it really easy to integrate web data into your products and agents.Here’s a demo video: https://www.tella.tv/video/build-faster-with-context-dev-api...Since it’s an API, here are the

Hy3

07/09 23:27

284 points · 63 comments

A possible future for Damn Interesting

07/09 23:25

167 points · 14 comments

Opinionated and Easy Pi.dev Configuration

07/09 23:19

83 points · 52 comments

TLS certificates for internal services done right

07/09 22:57

104 points · 69 comments

Why the Next Era of AI Is About Infrastructure, Not Just Models

07/09 22:50

21 points · 7 comments

No leap second will be introduced at the end of December 2026

07/09 22:16

190 points · 154 comments

Muse Spark 1.1

07/09 22:10

https://ai.meta.com/static-resource/muse-spark-1-1-evaluatio... [pdf]https://developer.meta.com/ai/resources/blog/build-with-muse...https://www.bloomberg.com/news/articles/2026-07-09/meta-star..., https://archi

The glass backbone: Why the Army's logistics will break in the next war

07/09 21:24

227 points · 292 comments

A road to Lisp: Why Lisp

07/09 21:06

60 points · 60 comments

Show HN: 18 Words

07/09 20:48

712 points · 260 comments

EU Parliament greenlights Chat Control 1.0

07/09 19:03

793 points · 393 comments

Show HN: Getting GLM 5.2 running on my slow computer

07/09 16:05

A few days ago I found myself trying out GLM 5.2 and was really positively impressed. The capabilities and security I was getting from this LLM are similar to those I've gotten from models like Claude or GPT, and this really surprised me.But then I thought, "I wonder how it would work on a

Postgres rewritten in Rust, now passing 100% of the Postgres regression tests

07/09 14:18

185 points · 236 comments

Show HN: I mapped 8.5M research papers into an interactive atlas

07/09 10:23

When I read papers, I have to jump between multiple tabs to find the dataset, code, videos, peer reviews, and so on. I tried to fix this with this project.It started as a project just for papers on arXiv, but after its initial success on Twitter (got like 1.9k views: the most I have gotten for a pos

Almost Always Unsigned

07/09 03:35

32 points · 45 comments

Girls just wanna have fast MPMC queues with bounded waiting

07/07 03:46

90 points · 16 comments

I Changed My Name

07/06 22:15

9 points · 0 comments

Buried Apple feature turns an iPhone into the perfect kids' dumb phone

07/06 18:53

https://archive.is/LV6Cw

How to Follow a Drummer

07/06 12:00

27 points · 21 comments

Train SIM Created by Just One Person Is Being Called the Best Ever Made

07/05 16:40

92 points · 31 comments

Meta reuses old RAM in new servers with custom bridge chip

07/04 03:27

270 points · 186 comments

v2.1.205

07/09 05:22

What's changed Added an auto mode rule that blocks tampering with session transcript files Fixed --json-schema silently producing unstructured output when the schema was invalid, and schemas using the format keyword being rejected Fixed a message sent while Claude was working being silently lost whe

中文介绍 Anthropic旗下的Claude Code项目发布了v2.1.205版本更新。此次更新主要增加了自动模式规则，旨在阻止对会话转录文件的篡改。同时，修复了当JSON Schema无效时静默生成非结构化输出的问题，以及含有“format”关键词的Schema被拒绝的问题。此外，还修复了Claude工作时发送的一条消息。

v2.1.204

07/08 08:27

What's changed Fixed hook events not streaming during SessionStart hooks in headless sessions, which could cause remote workers to be idle-reaped mid-hook

中文介绍 Anthropic的Claude Code发布了v2.1.204版本更新。此版本修复了一个问题：在无头会话（headless sessions）的SessionStart钩子（hooks）期间，钩子事件无法流式传输，这可能导致远程工作器在钩子执行中途被错误地回收。

v2.1.203

07/08 05:06

What's changed Added a warning when your login is about to expire, so you can re-authenticate before background sessions are interrupted Added a grey ⏸ badge to the footer when in manual permission mode, making the active mode always visible Added the session's additional working directories to MCP

中文介绍 Anthropic的Claude Code发布了v2.1.203版本更新。新版本主要进行了用户体验改进，包括添加了登录即将过期时的警告提示，以便用户能及时重新验证身份，避免后台会话中断。此外，在手动权限模式下，页脚会显示一个灰色暂停徽章，使用户能清晰看到当前模式状态。

v2.1.202

07/07 06:51

What's changed Added a "Dynamic workflow size" setting in /config for controlling how large Claude generally makes dynamic workflows (small/medium/large agent counts) — an advisory guideline, not an enforced cap Added workflow.run_id and workflow.name OpenTelemetry attributes to telemetry emitted by

中文介绍 Anthropic的Claude Code发布了v2.1.202版本更新。此次更新新增了在/config中配置「动态工作流大小」的选项，允许用户选择Claude生成工作流的代理数量规模（小型、中型或大型），这作为指导性设置。此外，新版本还为workflow.run_id和workflow.name添加了OpenTelemetry属性，以增强可观测性。

v2.1.201

07/04 07:50

What's changed Claude Sonnet 5 sessions no longer use the mid-conversation system role for harness reminders

v2.1.200

07/04 00:52

What's changed Changed AskUserQuestion dialogs to no longer auto-continue by default; opt into an idle timeout via /config Changed the "default" permission mode to "Manual" across the CLI, --help, VS Code, and JetBrains; --permission-mode manual and "defaultMode": "manual" are accepted alongside def

v2.1.199

07/03 07:35

What's changed Stacked slash-skill invocations like /skill-a /skill-b do XYZ now load all leading skills (up to 5), not just the first Fixed SSL certificate errors (TLS-inspecting proxies, missing NODE_EXTRA_CA_CERTS, expired certs) burning retries before showing actionable guidance — they now fail

v2.1.198

07/03 06:55

What's changed Subagents now run in the background by default, so Claude keeps working while they run and is notified when they finish (previously a gradual rollout) Claude in Chrome is now generally available Added background agent notifications in claude agents — sessions that need input or finish

v2.1.197

07/01 01:56

What's changed Introducing Claude Sonnet 5: now the default model in Claude Code, with a native 1M-token context window and promotional pricing of $2/$10 per Mtok through August 31. Update to version 2.1.197 for access. https://www.anthropic.com/news/claude-sonnet-5

v2.1.196

06/30 07:27

What's changed Added support for organization default models — admins set it in the org console; it shows as "Org default" (or "Role default") in /model when you haven't picked one yourself Added readable default names for sessions at start, making them easier to identify and message Added clickable