2026-07-18（内容获取于 07/18 04:47）

Agility Robotics plants its flag in Tesla’s backyard

TechCrunch · 07/18 04:19

Agility is opening a new training center for its Digit robots in Fremont, California.

中文介绍 Agility Robotics在加利福尼亚州弗里蒙特市开设了其Digit机器人的新培训中心。此举被视为该公司在特斯拉「后院」扎根，旨在进一步推广和应用其人形机器人产品，提升用户操作与维护能力。

Shark’s versatile ChillPill cooling system is back to its best price

The Verge · 07/18 04:15

You can clip the versatile cooling gadget, wear it, and use it as a desk fan. | Image: Shark Portable fans are one of the easiest ways to stay cool during the summer, and you don’t have to spend much to find a decent one. If you’re looking for something more versatile, though, Shark’s ChillPill Pers

中文介绍 Shark推出的多功能个人冷却系统ChillPill，目前价格已回落至历史最低点。这款便携式设备可夹持、佩戴或作为桌面风扇使用，为夏季降温提供多种便利，以其多功能性吸引消费者关注。

AI-driven memory crunch jolts India’s smartphone market

TechCrunch · 07/18 04:09

India's smartphone slowdown highlights how the AI boom is reshaping consumer electronics, from pricing and demand to corporate strategy.

中文介绍 印度智能手机市场正经历由AI驱动的内存供应紧张带来的增长放缓。这凸显了AI热潮如何从定价、需求到企业战略等多个层面，全面重塑消费电子产品领域。市场变化表明对高性能硬件的需求正在显著增加。

Google-backed satellites for wildfire detection launch as smoke chokes US, Canada

Ars Technica · 07/18 03:50

The FireSat program can spot wildfires that other satellites miss.

中文介绍 在美国和加拿大烟雾弥漫之际，由谷歌支持的FireSat野火探测卫星项目已成功发射。该项目旨在利用其独特技术，发现其他卫星可能遗漏的野火，从而提高早期预警能力，有助于更及时地应对森林火灾威胁。

Apple and Google ordered to purge ‘nudify’ apps from App Stores

TechCrunch · 07/18 03:49

In letters sent to Apple and Google, San Francisco City Attorney David Chiu said that both companies have long been aware that they are hosting apps in violation of state law.

中文介绍 旧金山市检察官David Chiu已致函苹果和谷歌，要求两家公司从其应用商店中删除「nudify」类应用程序。Chiu指出，两公司长期以来都知晓这些应用违反州法律，此举旨在打击非法内容，保护用户权益。

GPT-5.6 is deleting user files when given full access, and OpenAI says it shouldn't but did

The Decoder · 07/18 03:35

OpenAI's GPT-5.6 has accidentally wiped users' entire home directories in several cases, mostly in the unprotected "Full Access Mode." The model overwrites a temporary directory variable and carries out destructive actions on its own instead of asking for confirmation. OpenAI has announced extra saf

中文介绍 OpenAI的GPT-5.6模型在某些情况下意外删除了用户的整个主目录，尤其是在未受保护的「完全访问模式」下。该模型会覆盖临时目录变量，并自主执行破坏性操作，而非请求确认。OpenAI表示此问题不应发生，但确实已出现。

TikTok is testing an AI likeness detection tool

The Verge · 07/18 03:34

TikTok is starting to test an opt-in tool that scans for AI likenesses and lets creators report them to the company, as spotted by social media consultant Matt Navarra. The tool is initially being tested with "some" US creators, TikTok US spokesperson Zachary Kizer tells The Verge. YouTube has been

中文介绍 TikTok正针对部分美国创作者测试一项可选的AI肖像检测工具。该工具旨在扫描AI生成的肖像，并允许创作者向公司举报。TikTok美国发言人Zachary Kizer确认了此项测试，以应对AI合成内容带来的挑战，维护平台真实性。

Nuclear startup Valar Atomics in talks to raise new funding at $6B valuation

TechCrunch · 07/18 03:22

The potential deal highlights a growing trend of complex, multi-stage funding rounds that mask true entry prices.

中文介绍 核能初创公司Valar Atomics正就新一轮融资进行洽谈，预计估值将达到60亿美元。这笔潜在交易揭示了当前融资市场的一个趋势，即复杂的、多阶段的融资轮次可能掩盖真实的进入价格，使得早期投资者回报模式更加复杂。

The Pentagon's Space Development Agency hasn't moved as fast as anyone would like

Ars Technica · 07/18 03:19

"Missiles are being launched at the joint force every single day in [Operation] Epic Fury."

中文介绍 五角大楼的空间发展局（SDA）的进展速度未能达到预期。有评论指出，在「史诗之怒」行动中，联合作战部队每天都面临导弹威胁，这凸显了SDA在太空防御能力建设上需要更快步伐，以应对日益严峻的挑战。

Just like Deepseek, China's Kimi K3 is forcing Western AI labs to question their compute advantage

The Decoder · 07/18 03:16

Moonshot AI has released Kimi K3, a model that by early assessments matches Anthropic's Opus 4.8, built by a team of just 300 people. Even OpenAI strategist Dean W. Ball calls it "very good," but, of course, warns that a world dominated by open-weight models would amount to "AI communism." The relea

中文介绍 中国月之暗面发布的Kimi K3大模型，初步评估显示其性能与Anthropic的Opus 4.8相当，且由仅300人的团队开发。这引发西方AI实验室对其算力优势的质疑。OpenAI战略家Dean W. Ball称赞其「非常好」，但警告开源权重模型或导致「AI混乱」。

Pebble founder Eric Migicovsky says his 30-day warranty is all about trust

The Verge · 07/18 03:14

The Pebble Appstore is filled with stuff, but you don’t need much of it. | Photo by Amelia Holowaty Krales / The Verge Pebble founder Eric Migicovsky says buyers of its new e-paper smartwatches should know what they're signing up for and trust Pebble to make things right if they run into issues, des

中文介绍 Pebble创始人Eric Migicovsky表示，公司为其新推出的电子纸智能手表提供30天保修，这一政策的核心在于「信任」。他强调，购买者需了解产品特性，并相信Pebble在出现问题时能妥善解决，以此建立品牌与用户之间的互信关系。

Hegseth wants a "High-T" military; doctors call it a clinical minefield

Ars Technica · 07/18 02:53

"We're turning the clock back on rational healthcare."

中文介绍 Hegseth提出建立一支「高睾酮」（High-T）军队的构想，但医生们普遍认为这属于「临床雷区」，并批评此举「将理性医疗带回过去」。专家们担忧这种非科学的军事政策可能对士兵健康造成严重风险，引发医疗伦理争议。

Taco Bell iceberg lettuce identified as source of cyclosporiasis in 5 states

Ars Technica · 07/18 02:45

Don't eat Taco Bell lettuce in Indiana, Kentucky, Michigan, Ohio, or West Virginia.

中文介绍 Taco Bell餐厅的卷心生菜被确定为导致印第安纳、肯塔基、密歇根、俄亥俄和西弗吉尼亚州爆发环孢子虫病的感染源。卫生部门建议消费者在上述五个州避免食用Taco Bell的生菜产品，以防范潜在的健康风险。

Apple Music is getting a price hike

The Verge · 07/18 02:14

Apple Music is more expensive now. In the US, an individual plan now costs $11.99 per month, a $1 bump up from the previous $10.99 price. A family plan now costs $19.99 per month, up from $16.99, and a student plan costs $6.99 per month, up from $5.99. Apple, in a statement to Music Business Worldwi

中文介绍 Apple Music在美国市场提高了订阅价格。个人计划月费从10.99美元上调至11.99美元，家庭计划从16.99美元增至19.99美元，学生计划则从5.99美元涨至6.99美元。此次价格调整影响了各类用户。

I replaced my space heater and ceiling fan with one Dyson appliance

TechCrunch · 07/18 02:00

Designed for year-round comfort, the Dyson Hot+Cool HF1 combines quiet operation and simple controls with Dyson's signature bladeless design.

中文介绍 戴森（Dyson）推出了其Hot+Cool HF1多功能家电，旨在提供全年舒适的体验。该产品将传统空间加热器和吊扇的功能合二为一，具备静音运行、简洁控制以及戴森标志性的无叶片设计。用户可以用一台设备替代两种传统电器。

codecrafters-io/build-your-own-x

Markdown · ★ 527,205 · 🍴 49,894 · 📈 1,070 stars today

Master programming by recreating your favorite technologies from scratch.

中文介绍 build-your-own-x 提供了通过从零开始重新实现流行技术来掌握编程的实践路径。它鼓励学习者通过亲手构建诸如 Git、Docker、Redis 等系统，深入理解其底层原理和工作机制。该项目是为那些寻求通过项目实战巩固计算机科学基础、提升编程技能，并获得更深层次技术洞察力的开发者和自学者设计的。

PostHog/posthog

Python · ★ 36,150 · 🍴 3,010 · 📈 437 stars today

🦔 PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all

中文介绍 PostHog 是一个面向开发者的全栈产品分析与开发平台，旨在帮助团队构建“自驱动”产品。它提供了一系列工具，包括 AI 可观测性、用户行为分析、会话回放、功能开关 (feature flags)、A/B 实验、错误追踪和日志收集等。PostHog 帮助产品经理、开发者和数据分析师全面理解用户行为，快速迭代产品功能，并优化用户体验。

HenryNdubuaku/maths-cs-ai-compendium

TypeScript · ★ 6,554 · 🍴 807 · 📈 248 stars today

Become a cracked AI/ML Research Engineer

中文介绍 maths-cs-ai-compendium 是一个旨在帮助有志成为顶尖 AI/ML 研究工程师的综合性学习资料库。它系统性地整理了数学、计算机科学和人工智能领域的核心知识，为学习者提供了清晰的学习路径和必备的理论基础。用户可以通过此项目深入理解 AI/ML 算法背后的原理，构建扎实的知识体系，从而在 AI 研发领域取得突破。

Nutlope/hallmark

CSS · ★ 11,898 · 🍴 594 · 📈 1,486 stars today

Anti-AI-slop design skill for Claude Code, Cursor, and Codex.

中文介绍 hallmark 致力于提供“反AI废话”的设计技巧，专注于优化 Claude Code、Cursor 和 Codex 等AI工具生成的代码或设计。它旨在帮助开发者和设计师提升AI输出的质量，使其更具人性化、美观和实用性，避免生成平庸或低质量的内容。适用于需要将AI辅助成果融入高质量项目的工作者。

github/copilot-sdk

Java · ★ 9,773 · 🍴 1,327 · 📈 234 stars today

Multi-platform SDK for integrating GitHub Copilot Agent into apps and services

中文介绍 GitHub Copilot SDK 是一个多平台开发工具包，旨在帮助开发者将 GitHub Copilot Agent 的强大 AI 辅助编程能力集成到自己的应用程序和服务中。通过此 SDK，各种开发环境、IDE 或自定义工具能够无缝接入 Copilot 的代码建议、自动补全等功能，从而为用户提供更智能、更高效的编程体验。它赋能开发者在更广泛的场景中利用 AI 提升生产力。

anthropics/cwc-workshops

TypeScript · ★ 1,558 · 🍴 489 · 📈 37 stars today

中文介绍 此项目可能包含Anthropic公司提供的，关于其AI模型（如Claude）或其核心理念——Constitutional AI（CWC）相关的教程和研讨会资料。它旨在帮助开发者和研究人员深入理解和应用Anthropic的AI技术，特别是在AI安全、伦理和模型行为控制方面的实践方法和指南。

PrismML-Eng/Bonsai-demo

Shell · ★ 1,695 · 🍴 166 · 📈 279 stars today

Bonsai Demo

中文介绍 PrismML-Eng/Bonsai-demo 是一个机器学习工程相关的演示项目。尽管描述简短，但结合项目所有者“PrismML-Eng”，可以推断该项目可能展示了某种 ML 工程实践、模型训练流程或特定的 AI 应用场景。它适合对机器学习工程实践感兴趣的开发者和研究人员探索。

protocolbuffers/protobuf

C++ · ★ 71,530 · 🍴 16,191 · 📈 18 stars today

Protocol Buffers - Google's data interchange format

中文介绍 Protocol Buffers（Protobuf）是Google开发的一种高效、语言中立、平台无关的结构化数据序列化机制。它允许开发者定义数据结构，然后通过编译生成多种编程语言的代码，实现数据的快速序列化、反序列化和传输。Protobuf常用于微服务通信、数据存储和API接口定义，性能通常优于XML和JSON。

tirth8205/code-review-graph

Python · ★ 19,699 · 🍴 2,106 · 📈 57 stars today

Local-first code intelligence graph for MCP and CLI. Builds a persistent map of your codebase so AI coding tools read only what matters, with benchmarked context reductions on reviews and large-repo workflows.

中文介绍 Code Review Graph是一个本地优先的代码智能图谱工具，专为MCP（多组件项目）和CLI环境设计。它构建了一个持久化的代码库映射，旨在优化AI编码工具的上下文理解，确保AI在代码审查和处理大型仓库时仅读取关键信息。通过大幅减少AI上下文长度，提升了代码理解效率和审查质量。

docusealco/docuseal

Ruby · ★ 17,789 · 🍴 1,768 · 📈 152 stars today

Open source DocuSign alternative. Create, fill, and sign digital documents ✍️

中文介绍 DocuSeal是一个开源的DocuSign替代方案，旨在提供一套完整的数字文档处理解决方案。用户可以通过它创建自定义的电子文档模板，在线填写表单，并进行合法的数字签名。它适用于个人、团队或企业，用于合同签署、表单提交、协议审批等场景，实现无纸化办公和高效的文档流转。

openinterpreter/openinterpreter

Rust · ★ 66,294 · 🍴 5,699 · 📈 431 stars today

A coding agent for open models like Kimi K3

中文介绍 openinterpreter 是一个专为低成本 AI 模型设计的编程智能体。它通过提供一个代码执行环境，赋能经济型 AI 模型也能高效完成复杂的编程任务。该项目旨在降低开发者在利用 AI 辅助编程时对昂贵大型模型的需求，使 AI 编程自动化更具成本效益。用户可以是需要自动化代码生成、调试或项目管理的开发者，特别适合在资源有限的环境下，利用较小规模的语言模型提升开发效率。

RyanCodrai/turbovec

Python · ★ 13,253 · 🍴 1,177 · 📈 280 stars today

A vector index built on TurboQuant, written in Rust with Python bindings

中文介绍 TurboVec是一个高性能的向量索引库，基于TurboQuant技术构建，并使用Rust语言编写，同时提供了便捷的Python绑定。它旨在提供高效的向量存储和检索能力，支持快速的相似性搜索，适用于RAG（检索增强生成）、推荐系统、语义搜索等需要处理大量高维向量数据的场景。Rust的底层性能确保了查询速度，Python绑定则方便了机器学习和数据科学领域的用户。

HKUDS/DeepTutor

Python · ★ 27,290 · 🍴 3,642 · 📈 528 stars today

DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.

中文介绍 DeepTutor 是一个提供“终身个性化辅导”的智能教育系统。它利用深度学习等先进技术，为学习者提供定制化的、贯穿整个学习过程的辅导服务。该项目旨在解决传统教育中缺乏个性化指导的问题，通过适应学生的学习进度、风格和知识盲点，提供实时反馈和定制内容。它非常适合学生进行自主学习，教师辅助教学，以及任何寻求高效、个性化学习体验的终身学习者。

OpenCut-app/OpenCut

TypeScript · ★ 74,737 · 🍴 7,540 · 📈 1,077 stars today

The open-source CapCut alternative

中文介绍 OpenCut 是一个开源的视频剪辑软件，旨在成为流行剪辑工具 CapCut 的免费替代品。它为视频创作者和内容编辑提供了一个功能齐全且可定制的平台，用于处理视频素材。用户可以利用其免费开源的特性，在不依赖商业软件的情况下进行高效的视频制作和后期编辑。

You just hired a million bad employees.

@gsivulka · 10.2K 粉丝 · 591.2K 阅 · 512 赞 · 99 转

07/15 00:41

AI was supposed to replace human labor. It did the opposite. For the first time in history, humans are cheaper than software. And AI is creating more jobs than it eliminates. Technology has

中文介绍 博主gsivulka以「你刚雇佣了一百万个糟糕的员工」为引子，暗示了对AI工具或其应用效果的批判性观点。

Making Fable Cheaper Than Opus

@joon_h_lee · 515 粉丝 · 366.3K 阅 · 521 赞 · 43 转

07/14 01:03

We replaced Opus 4.8 with Fable 5, and Devin’s bill went down. Fable 5 costs twice as much per token as Opus 4.8. But when we ran both models on FrontierCode 1.1 using our new Fusion architecture,

中文介绍 博主分享了通过用Fable 5替换Opus 4.8显著降低Devin账单的经验。尽管Fable 5的每token成本是Opus 4.8的两倍，但利用其新Fusion架构在FrontierCode 1.1上运行模型后，整体费用反而下降。这表明通过优化模型架构和运行时环境，即使使用成本更高的模型也能实现更优的经济效益。

The Most Human Technology Ever Made

@illscience · 30.4K 粉丝 · 332.5K 阅 · 539 赞 · 69 转

07/14 01:05

Over the weekend, Mira Murati wrote a phenomenal essay that you should read in full: “The future worth building is human.” I think it’s one of the best essays of the year, because it gets at the

中文介绍 博主推荐OpenAI首席技术官Mira Murati撰写的文章「值得建设的未来是人性化的」。该文章探讨了将人类价值融入AI发展的未来方向，并被博主评价为年度最佳文章之一。它强调了在AGI时代，技术应以人为本，而非单纯追求技术突破。

The Great Wave Has Arrived (from GLM CEO Jie Tang)

@bingxu_ · 6.5K 粉丝 · 318.6K 阅 · 505 赞 · 88 转

07/11 23:10

-- Bing Xu's Note --- I came across an internal GLM letter on the Chinese app RedNote, purportedly written by @jietang, and translated the Chinese text in the images into English. --- Letter Start

中文介绍 智谱AI CEO唐杰发表了「大模型浪潮已至」的观点，指出了当前人工智能发展的重要转折点，并对未来AI大模型的趋势与影响进行了预判。

GLM 5.2 With Vision

@part_harry_ · 1.5K 粉丝 · 225.6K 阅 · 503 赞 · 46 转

07/16 12:24

GLM 5.2 is one of the best currently available open source language models. However, unlike other flagship models like Qwen, Kimi and Minimax, GLM 5.2 does not support image inputs. We took this as a

中文介绍 博主讨论开源语言模型GLM 5.2的视觉能力（Vision），可能是在为其探索或开发图像输入功能。这一举措旨在增强GLM 5.2的多模态处理能力，使其能够处理视觉信息，从而拓宽应用范围并提升其在与Qwen、Kimi等模型竞争时的竞争力。

The Self-Driving Company

@amasad · 472.6K 粉丝 · 223.7K 阅 · 560 赞 · 65 转

07/17 01:07

We are beginning to see what happens when a company learns to operate itself. In the past six months, engineers at Replit have nearly tripled code output. Review times held steady. Reversions and

中文介绍 博主提出了「自驱动公司」这一概念，暗示企业运营未来将高度自动化，能像自动驾驶汽车一样自我管理和优化。这可能涉及利用AI和自动化技术，大幅提升公司的运营效率、决策速度和资源配置能力，预示着企业组织形态和生产力的一次重大变革。

How OpenAI’s Sol Finally Learned Design Taste

@DesignArena · 17.6K 粉丝 · 191.5K 阅 · 509 赞 · 46 转

07/16 00:36

We benchmarked GPT-5.6 Sol on Design Arena’s Web Design (Non-Agentic) Arena, and we were surprised to find that it ranks 1st overall. This is 18 places higher than its predecessor GPT-5.5, and is the

中文介绍 DesignArena 提出了「How OpenAI’s Sol Finally Learned Design Taste」的问题，可能旨在探讨 OpenAI 的 Sol 模型在学习和掌握设计美学方面的进展和方法。

The Reverse Information Paradox

@satyanadella · 7.0M 粉丝 · 146.5K 阅 · 1.1K 赞 · 162 转

07/12 23:09

In the age of intelligence, how should firms protect their core IP? Nobel Prize winning economist Kenneth Arrow famously described a paradox in the market for information. “Its value for the

中文介绍 微软CEO Satya Nadella探讨了「智能时代」企业如何有效保护核心知识产权的挑战。他引用诺贝尔经济学家Kenneth Arrow的信息悖论，深入分析了AI背景下知识价值与保护机制的新思考。

The Next Horizon in Agents

@btaylor · 158.6K 粉丝 · 144.3K 阅 · 511 赞 · 40 转

07/16 22:48

Introducing Horizon Today, Sierra is announcing Horizon, a platform that enables agents to pursue long-horizon goals, like originating a loan or getting prior authorization for a healthcare procedure.

中文介绍 Sierra公司发布了名为Horizon的新平台，旨在赋能AI Agent处理复杂且「长周期」的实际业务目标。该平台支持Agent完成如发起贷款或获取医疗程序预授权等多步骤任务。这标志着AI Agent从执行短期指令向处理更深层次、更具商业价值的复杂流程迈进，为企业自动化提供了新的可能性。

Gemini API Managed Agents: free tier, triggers & more

@GoogleAIStudio · 186.6K 粉丝 · 111.0K 阅 · 512 赞 · 51 转

07/17 01:05

Today we’re announcing new capabilities for Managed Agents in Gemini API, including free tier availability, budget control guardrails, and scheduled triggers. Building on our previous release of

中文介绍 Google AI Studio宣布Gemini API的Managed Agents新增功能，包括提供免费层级、预算控制保护措施以及支持定时触发器。这些更新建立在之前的发布基础上，旨在降低开发者使用门槛，赋予他们更大的灵活性来管理和部署AI Agent。新功能将促进更广泛的Agent应用，并帮助用户更好地控制成本与调度。

ORACLE: Official AI Agents Trade on Polymarket

@Oracle__Promo · 4.1K 粉丝 · 106.5K 阅 · 1.8K 赞 · 543 转

07/11 07:04

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 ORACLE 项目宣布，到 2026 年，自主 AI Agent 已成为预测市场 Polymarket 上最有效的策略之一。目前 Polymarket 超过 30% 的交易活动来自算法和 AI 驱动的钱包。这篇推文预示了 AI Agent 在去中心化预测市场中的巨大潜力和应用前景，可能涉及产品发布或愿景分享。

ORACLE: Official AI Agents Trade on Polymarket

@Oracle__Trade · 4.0K 粉丝 · 105.3K 阅 · 1.9K 赞 · 557 转

07/11 07:03

In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We

中文介绍 ORACLE 项目宣布，到 2026 年，自主 AI Agent 已成为预测市场 Polymarket 上最有效的策略之一。目前 Polymarket 超过 30% 的交易活动来自算法和 AI 驱动的钱包。这篇推文预示了 AI Agent 在去中心化预测市场中的巨大潜力和应用前景，可能涉及产品发布或愿景分享。

A Framework for Frontier AI and the Dawning of a New Age

@demishassabis · 1.4M 粉丝 · 98.1K 阅 · 1.5K 赞 · 236 转

07/14 17:10

This is a pivotal moment in human history. Artificial General Intelligence (AGI), a system that exhibits all the cognitive capabilities the brain has, is probably only a few short years away. When we

中文介绍 Google AI CEO Demis Hassabis认为，人类正处于关键历史时刻，通用人工智能（AGI）的实现可能仅在数年之内。AGI系统有望展现人类大脑的全部认知能力。他可能在分享一个关于前沿AI的框架，以应对即将到来的新时代。

gpt-5.6-sol without hitting limits

@theo · 356.6K 粉丝 · 77.0K 阅 · 774 赞 · 27 转

07/12 06:59

I've burned over $200,000 of tokens with gpt-5.6-sol. It's a great model. That said, it's a bit too easy to hit your limits on the $200 Codex Pro sub. OpenAI has been generous about resets, but that

中文介绍 博主Theo分享了关于在应用中使用模型「gpt-5.6-sol」的经验，侧重于如何避免触及其固有的使用限制。这篇帖子可能提供了在特定场景下，实现模型无障碍运行的策略或操作建议。

How we built our knowledge base

@cerebras · 60.4K 粉丝 · 70.7K 阅 · 509 赞 · 38 转

07/17 02:27

Authors: @hi_im_isaac_, @learnwdaniel, @gaozenghao note: the interactive version of full technical blog available: https://www.cerebras.ai/blog/how-we-built-our-knowledge-base Employees ask our internal knowledge base more than 15,000

中文介绍 Cerebras发布技术博客，详细阐述了他们如何构建企业内部知识库。该知识库每月处理员工超过15,000次的查询，极大提高了工作效率。博文由多位作者共同撰写，提供了实际案例和技术细节，为其他组织构建高效智能知识管理系统提供了宝贵的实践经验和设计思路。

Kimi K3 on KernelBench

@elliotarledge · 40.7K 粉丝 · 69.3K 阅 · 506 赞 · 45 转

07/17 17:24

This is the Kimi K3 post you guys have been waiting for. I got some early access to this model and have been testing it on kernels, and even before seeing the benchmark scores I was impressed with its

中文介绍 博主获得了Kimi K3模型的早期访问权限，并对其进行了初步测试和KernelBench基准评估。在未查看具体分数前，博主已对Kimi K3的性能印象深刻。这篇分享为期待Kimi K3的用户提供了第一手体验和性能洞察，预示着该模型可能在某些任务上展现出强大的能力。

Hierarchical Denoising For Multi-Step Visual Reasoning

👍 1

07/16 08:00

Video models are evolving into vision foundation models, yet they still lack human-like multi-step reasoning. Streaming autoregressive diffusion models are efficient but limited in reasoning, while bidirectional diffusion enables global revision with high inference costs due to dense frame-level den

中文介绍 视频模型正发展为视觉基础模型，但在多步视觉推理方面仍欠缺类人能力。该研究探讨了流式自回归扩散模型和双向扩散模型的优缺点，前者高效但推理有限，后者支持全局修正但推理成本高昂。论文提出一种分层去噪方法，旨在提升视频模型的复杂视觉推理能力，以克服现有模型的局限性。

Partition, Prompt, Aggregate: Statistical Self-Consistency in Language Models

👍 4

07/16 08:00

In-context learning is commonly interpreted as a form of conditional inference, in which the prompt specifies a context and the model's output is treated as an estimate of the corresponding conditional distribution. If this interpretation holds, then LLM estimates should satisfy basic probabilistic

中文介绍 上下文学习常被视为一种条件推理形式，其中提示词设定语境，模型输出则被视为对应条件分布的估计。该论文提出「Partition, Prompt, Aggregate」方法，旨在探讨并提升大型语言模型（LLMs）估计的统计自洽性。研究表明，如果这种解释成立，LLM的估计应满足特定的统计特性，从而提高模型输出的可靠性。

RoboTTT: Context Scaling for Robot Policies

👍 12

07/16 08:00

Recent robot foundation models operate with single-step or short-history visuomotor context. We introduce Test-Time-Training Robot Policies (RoboTTT), a robot model and training recipe that scale visuomotor context to 8K timesteps, three orders of magnitude beyond state-of-the-art policies, without

中文介绍 现有的机器人基础模型通常仅处理单步或短期视觉运动上下文。该论文提出「RoboTTT」（测试时训练机器人策略），这是一种机器人模型和训练方法，能将视觉运动上下文扩展到8K时间步。这比现有技术水平提升了三个数量级，有望显著增强机器人策略的长期规划和决策能力。

MeanFlowNFT: Bringing Forward-Process RL to Average-Velocity Generators

👍 9

07/16 08:00

MeanFlow generators achieve fast few-step sampling by predicting average velocities over time intervals, making them attractive for efficient generation. Reinforcement learning (RL) has become a powerful way to align diffusion and flow models with human preferences and task-specific objectives. In p

中文介绍 MeanFlow生成器通过预测时间间隔内的平均速度，实现快速的少步采样，适用于高效生成任务。强化学习（RL）已成为使扩散模型和流模型与人类偏好及特定任务对齐的有效方法。该论文提出了「MeanFlowNFT」，将前向过程强化学习引入平均速度生成器，旨在进一步优化生成模型的性能和用户体验。

SearchOS-V1: Towards Robust Open-Domain Information-Seeking Agent Collaboration

👍 48

07/16 08:00

Recent advances in Tool-Integrated Large Language Models have made web search a core capability of information-seeking agents. However, as interaction histories grow, agents increasingly struggle to track task progress. When search attempts fail to yield useful evidence, current single- and multi-ag

中文介绍 随着工具集成大型语言模型的发展，网络搜索已成为信息搜寻代理的核心能力。然而，交互历史的增长导致代理难以跟踪任务进度，尤其当搜索未能提供有效证据时。该论文提出「SearchOS-V1」，旨在构建更鲁棒的开放领域信息搜寻代理协作系统，以解决现有代理在处理复杂和失败搜索场景中的局限性。

HoloGeo: Mitigating Landmark Bias in Geo-localization via Evidence-Driven Reasoning

👍 0

07/16 08:00

Recent advances in Vision-Language Models (VLMs) have significantly improved image geo-localization, yet existing models remain susceptible to landmark bias, causing them to overlook geographical cues or form spurious correlations, ultimately resulting in inaccurate localization. To systematically i

中文介绍 视觉语言模型（VLMs）显著提升了图像地理定位能力，但现有模型易受地标偏见影响，导致忽视地理线索或形成虚假关联，最终定位不准确。该论文提出「HoloGeo」方法，旨在通过证据驱动推理减轻这种地标偏见。通过更有效地整合视觉与语言信息，HoloGeo有望提升地理定位的鲁棒性和准确性。

BadWAM: When World-Action Models Dream Right but Act Wrong

👍 36

07/16 08:00

World-action models (WAMs) are emerging as a promising foundation for embodied control: rather than predicting actions alone, they learn representations that couple action generation with future world prediction. This coupling is often viewed as a source of robustness, interpretability, and safety,

中文介绍 世界-行动模型（WAMs）通过将行动生成与未来世界预测相结合，成为具身控制的潜在基础。这种耦合通常被视为鲁棒性和可解释性的来源。然而，该论文提出的「BadWAM」研究了WAMs在“正确预测世界却错误执行行动”的情况。文章分析了导致这种偏差的原因，并探讨了提升模型行动准确性的策略。

MM-IssueLoc: A Controlled Benchmark for Evaluating Visual Evidence in Multimodal Repository-Level Issue Localization

👍 0

07/16 08:00

Real repository issues routinely include visual evidence such as screenshots, error dialogs, rendered UI states, and logs, yet repository-level issue localization is evaluated mostly as a text-only task. Existing multimodal SE benchmarks evaluate end-to-end repair, entangling localization with patch

中文介绍 实际软件仓库问题常包含截图、错误对话框等视觉证据，但仓库级问题定位主要仍是文本任务。现有多模态软件工程基准侧重端到端修复，而非纯粹定位。该论文提出「MM-IssueLoc」，一个受控基准测试集，用于专门评估多模态仓库级问题定位中视觉证据的作用。它旨在推动更准确、更全面的问题分析。

Setup Complete, Now You Are Compromised: Weaponizing Setup Instructions Against AI Coding Agents

👍 0

07/16 08:00

AI coding agents set up projects by reading documentation and installing the dependencies it lists, without verifying their names, sources, or known vulnerabilities. By editing only a README, requirements file, or Makefile, an attacker can redirect the agent to an untrusted registry, a known-vulnera

中文介绍 AI编程代理通过阅读文档安装依赖项，但缺乏对名称、来源或已知漏洞的验证。攻击者可通过修改README文件、依赖列表或Makefile，将代理重定向到不受信任的注册表，从而导致系统被入侵。该论文揭示了针对AI编程代理的潜在供应链攻击风险，并强调了加强其安装过程安全性的必要性。

SUFLECA: Scaling Up Feature Learning for CAD-to-image Alignment

👍 3

07/16 08:00

CAD-to-image alignment aims to estimate an object's 9D pose (rotation, translation, and anisotropic scale) from a single RGB image, enabling applications in robotics and augmented reality. Recent zero-shot methods use visual foundation models to match image regions to CAD models, yet typically their

中文介绍 CAD-到图像对齐技术旨在从单张RGB图像中估计物体9D姿态（旋转、平移和各向异性尺度），在机器人和增强现实中具有应用。近期零样本方法利用视觉基础模型进行图像区域与CAD模型匹配。该论文提出「SUFLECA」方法，旨在扩展特征学习规模，以提升CAD-到图像对齐任务的准确性和鲁棒性。

Video = World + Event Stream

👍 14

07/16 08:00

We present Wan-Streamer v0.3, which reframes our native-streaming interaction model under a single organizing view: a video is a world plus an event stream. The world is the persistent context in which a video unfolds, including the environment, scene, subjects, ambient acoustic conditions, voice ch

中文介绍 该论文推出「Wan-Streamer v0.3」，提出一种将视频重新定义为“世界+事件流”的交互模型。其中，“世界”代表视频展开的持久性上下文，涵盖环境、场景、主体和环境声学等；“事件流”则捕捉动态变化。这一新框架旨在为本地流媒体交互提供统一视图，深化对视频内容的理解和处理。

AeroAct: Action-Centered World-Action Models for Language-Conditioned Quadrotor Flight

👍 0

07/16 08:00

Language-conditioned quadrotor flight requires a policy to ground semantic goals, anticipate the visual consequences of ego-motion, and output control references that remain smooth and dynamically executable under rapidly changing first-person views. Existing aerial vision-language navigation and vi

中文介绍 语言条件下的四旋翼飞行要求策略能实现语义目标，预测自我运动的视觉后果，并在快速变化的第一人称视角下输出平滑且动态可执行的控制指令。现有空中视觉语言模型在这方面表现不足。该论文提出「AeroAct」，一种以行动为中心的世界-行动模型，旨在解决语言控制四旋翼飞行的复杂挑战，提升其自主导航和操作能力。

From Draft to Draft-Free: One-Step Video Object Removal via Privileged Distillation and Fast Planting

👍 0

07/16 08:00

Video object removal is a fundamental yet challenging task in video editing. Despite recent progress, existing methods typically fall into two categories. Traditional approaches based on optical flow or attention mechanisms often introduce noticeable artifacts and yield unnatural results. In contras

LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU Budget

👍 39

07/16 08:00

A growing gap separates inference context lengths from RL post-training: inference systems are approaching million-token contexts, while post-training workloads often remain at 256K tokens or below and rely on length generalization at deployment. The gap is especially important for AI agents, whose

Steering Robustness into World Action Models via Mechanistic Interpretability and Optimal Control

👍 0

07/16 08:00

World Action Models (WAMs) enable semantically- and physically-informed control but are brittle under distribution shift. In this work, we use mechanistic interpretability to study how robustness-relevant perturbations are represented in WAM activation space. Comparing activations across successful

VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understanding

👍 107

07/16 08:00

Recent advances in video understanding have spanned motion, long video, and streaming interaction, driving this field toward real-world applications. Despite this progress, current open-source models remain limited in several ways. They often struggle to generalize across diverse video types, making

StructureClaw: Traceable LLM Agents and an Executable Benchmark for Structural Engineering Workflows

👍 0

07/16 08:00

Addressing a structural-engineering request requires more than a single answer; it requires a chain of interdependent artifacts: interpreted requirements, a computable model, validation records, solver outputs, code-check records, and a final report. Evaluations centered on question answering or scr

Proof-or-Stop: Don't Trust the Agent, Trust the Evidence -- Loop Engineering for Verifiable Evidence-Gated Lifecycle Control

👍 0

07/16 08:00

Autonomous coding agents increasingly execute multi-step software work, but lifecycle states such as reviewed, tested, DONE, and ready-to-merge remain claims unless supported by current evidence. We present Proof-or-Stop Lifecycle Control, a method that permits lifecycle transitions only when fresh,

Towards Human-like Physical Intelligence: LifelongVision-Language-Action Learning for Robotic Manipulation

👍 0

07/16 08:00

Similar to the natural capabilities of humans to sequentially learn new tasks, robots with Vision-Language-Action (VLA) models should possess lifelong learning ability to learn a new task when deployed in open-world environments. However, most recently proposed lifelong learning models aim to effect

SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning

👍 65

07/16 08:00

Large language models are increasingly trained as interactive agents for long-horizon tasks involving multi-turn interaction, tool use, and environment feedback. Outcome-based reinforcement learning (RL) provides a practical optimization paradigm, but its sparse trajectory-level rewards offer limite

WanSong v1.0 Technical Report

👍 8

07/16 08:00

Music generation foundation models have recently attracted significant industry attention. However, achieving efficient generation and high-fidelity long-form audio while supporting controllability remains challenging. To address these needs, we present WanSong, a simple yet powerful approach for lo

Does Multi-Agent Debate Improve AI Feedback on Research Papers?

👍 0

07/16 08:00

Probably not, at least for meta-analyses in economics. In a pre-registered, identity-masked, within-paper experiment, the authors of 44 meta-analyses ranked three AI reports on their own paper by usefulness for improving it: a single pass by a frontier model against two multi-agent debate tools we b

Reflex: Real-Time VLA Control through Streaming Inference

👍 0

07/16 08:00

Flow matching Vision-Language-Action (VLA) models promise precise continuous control, but their iterative denoising nature introduces fundamental incompatibilities with real-time robotics: global timestep injection invalidates KV-caching, forcing a choice between slow O(N^2) re-computation or mathem

Project Kaleidoscope: Contextual, Human-Aligned Evaluation for Real-World AI Applications

👍 0

07/16 08:00

Evaluations (Evals) are a deployment bottleneck for real-world AI applications: public benchmarks rarely match a team's users, context, or policies, and human review is often tedious to scale. Motivated by our work with AI applications in the public sector, this project addresses recurring evaluatio

VIABench: A Comprehensive Video Benchmark Collected from Blind Individuals for Visual Impairment Assistance

👍 7

07/16 08:00

Visually impaired individuals (VIIs) encounter significant daily challenges due to limited access to visual information. Although Multimodal Large Language Models (MLLMs) have achieved impressive results on general vision and language tasks, their practical utility in real-world blind assistance sti

MCPEvol-Bench: Benchmarking LLM Agent Performance Across Dynamic Evolutions of MCP Servers

👍 0

07/16 08:00

As Model Context Protocol (MCP) servers emerge as the core infrastructure for connecting LLMs with external tools, existing benchmarks leverage real-world MCP servers to evaluate LLM agents' tool-using capabilities. However, these benchmarks overlook the continuous evolution of tool interfaces and f

Action QFormer: Structured Representation Shaping under Action Supervision in Vision-Language-Action Models

👍 0

07/16 08:00

Action supervision in vision-language-action (VLA) models is often treated as a downstream objective for learning action prediction. In this paper, we study it instead as a force that shapes inherited multimodal representations. We show that this shaping has a dual effect: it is necessary for formin

Bad Memory: Evaluating Prompt Injection Risks from Memory in Agentic Systems

👍 0

07/16 08:00

A growing class of agentic systems maintain persistent state across sessions through memory files, behavioral preferences, and knowledge bases. While this makes agents more useful and self-improving, it also creates a new attack surface for prompt injections in which malicious instructions can be em

Representation-Aligned Tactile Grounding for Contact-Rich Robotic Manipulation

👍 0

07/16 08:00

Tactile-enhanced vision-language-action (VLA) policies have been introduced for contact-rich manipulation, where critical interaction states are often hidden from vision. Future tactile prediction is a promising way to use touch because it turns tactile outcomes into supervision for action-induced c

SoftNav: Injecting 3D Scene Tokens into VLMs for Embodied Navigation

👍 0

07/16 08:00

In goal-directed embodied navigation, where an agent must locate a specified target in an unseen environment, 3D scene understanding and navigation reasoning must work in concert. Current approaches transmit 3D scene information to vision-language models (VLMs) through text, suggesting a representat

Unabyss for Claude

07/18 04:37

Shared memory across all apps and LLMs. In Claude

中文介绍 Unabyss for Claude是一款为Claude设计的产品，它旨在实现所有应用程序和大型语言模型（LLM）之间的共享内存功能。用户可以通过该工具，在不同应用和LLM之间高效地共享信息和上下文，提升工作流程的连贯性。

Timely

07/18 04:29

Pull your calendar availability in 3 seconds

中文介绍 Timely是一款日程管理工具，旨在帮助用户在短短3秒钟内快速获取并分享其日历可用时间。这款产品简化了会议安排和协作过程，使用户能够更高效地协调日程，节省沟通时间。

Yapper Leaderboard

07/18 04:28

See the biggest startup yappers on X/Twitter

中文介绍 Yapper Leaderboard是一款社交媒体分析工具，专注于识别和展示在X（原Twitter）平台上关于初创公司讨论最活跃的用户。该产品通过排行榜形式，帮助用户发现并跟踪那些在社交媒体上对初创企业发表见解和评论最多的「发声者」。

PixyCAD

07/18 04:28

Fast & precise 3D CAD built natively for iPad and Mac

中文介绍 PixyCAD是一款专为iPad和Mac平台原生开发的快速且精确的3D实体建模CAD软件。该产品旨在提供高效的设计体验，充分利用Apple生态系统的优势，帮助用户在移动和桌面设备上进行专业的3D设计工作。

Aye

07/18 04:28

Your teachable AI intern for everyday browser work

中文介绍 Aye是一款专为日常浏览器工作设计的“可教导”AI实习生。它能够学习并适应用户的工作习惯，在浏览器环境中提供智能辅助，帮助用户更高效地完成网页浏览、信息处理等任务，从而提升日常工作效率。

Pebbles Ai

07/18 04:28

AI sales platform for modern B2B teams

中文介绍 Pebbles Ai是一个面向现代B2B团队的人工智能销售平台。该平台利用AI技术优化销售流程，旨在帮助企业客户实现更高效的销售策略、客户关系管理和业绩增长，从而提升B2B销售团队的整体竞争力。

Paradigm

07/18 04:28

Turn any goal into a personalized, adaptive learning path.

中文介绍 Paradigm是一款创新产品，旨在帮助用户将任何目标转化为个性化、自适应的学习路径。它通过智能算法，根据用户的具体情况和学习进度，自动调整和优化学习内容及节奏。这有助于用户更高效地达成目标，无论是技能提升还是知识获取，都能获得定制化的学习体验。

Nuvio

07/18 04:27

Connect your X bio to your MRR and GA4 for auto updates

中文介绍 Nuvio是一款数据连接工具，可将用户的X（原Twitter）个人简介与月经常性收入（MRR）及Google Analytics 4（GA4）数据进行关联，实现自动更新。该产品旨在帮助用户，特别是创始人或营销人员，在其社交媒体资料上实时展示关键业务指标。

Pocket Screen

07/18 04:22

Keep any Mac window visible in a floating mini screen

中文介绍 Pocket Screen是Toybird Labs开发的一款Mac应用程序，其核心功能是允许用户将任何Mac窗口以浮动迷你屏幕的形式保持可见。该工具旨在提升多任务处理效率，使用户在执行其他操作时，仍能轻松监控或参考特定窗口的内容。

Kimi K3

07/18 04:21

The world's first open 3T-class model

中文介绍 Kimi K3，作为Kimi AI助手的核心模型，号称是全球首个开放的3T级别模型。这一创新性的大模型，有望为人工智能领域带来新的突破，特别是在处理大规模数据和复杂任务方面提供强大的能力，并可能推动相关技术的开放生态发展。

Devin AI: The Full Beginner’s Guide (Better Than Claude Code?)

07/17 02:25

中文介绍 这则YouTube视频提供了Devin AI的全面初学者指南，深入探讨了这款人工智能工具在代码开发方面的应用潜力。视频可能通过演示和比较，向观众展示Devin AI的功能和使用方法，并讨论其与Claude Code相比的优劣，以帮助初学者快速掌握。

Director by OpenArt

07/17 02:12

中文介绍 这则YouTube短视频介绍了OpenArt公司推出的一款名为「Director」的新产品或功能。该视频可能简要展示了「Director」的主要特点和用途，暗示其可能是一款用于创意或艺术领域的人工智能工具。

Codex NEW Browser Just Destroyed Chrome

07/15 21:00

中文介绍 Codex和Claude两大AI模型近期发布了浏览器更新，预计将对用户与AI的互动方式产生显著影响。这些更新可能包括功能改进、性能优化或更深度的浏览器集成，旨在提升用户体验并拓宽AI工具的应用场景。

OpenAI Just Merged ChatGPT and Codex. This Changes Everything.

07/13 01:24

Grok 4.5 + Cursor: Full Guide

07/10 01:51

Build Beautiful Websites with Claude Fable 5 (Full Guide)

07/07 01:37

How to Create Your Own 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

还有 2 天， fable5 就要计费了，各位要转向 codex 吗？

07/17 21:44

6 回复 · 程序员 节点

Kimi k3， gpt5.6， gemini3.1 生成的单页面哪个好看

07/17 20:14

15 回复 · 程序员 节点

强烈推荐一款性价比极高的 Linux 本搞 vibe coding

07/17 12:24

28 回复 · Linux 节点

国行 iPhone 开始出现苹果 AI 流量连接

07/17 11:51

13 回复 · Apple 节点

如果以后 Apple AI 国行版上线，国行的 iPhone 可以在科学上网前提下，使用国外的 Apple AI 吗？

07/16 16:24

16 回复 · Apple 节点

已故妻子的 iPhone 因失窃设备保护卡在 Face ID 验证，已知锁屏及 Apple 账户密码，能否提取微信数据？

07/16 16:07

99 回复 · Apple 节点

如何给非大陆版 AirPod Pro 购买美版 Apple Care+

07/16 14:56

7 回复 · Apple 节点

[Mac] 自荐我的新 App： DropPocket - 屏幕顶部的临时文件口袋，限时特价 $3.99 🎉

07/16 10:45

6 回复 · Apple 节点

iOS27 太流畅了，垃圾 26

07/16 09:39

15 回复 · Apple 节点

Notability App 美区买断 40% off, 仅限三天

07/16 04:33

11 回复 · Apple 节点

Flock CEO Apologizes for Calling Activists 'Terrorists'

07/18 03:53

32 points · 12 comments

The Zilog Z80 has turned 50

07/18 03:41

44 points · 6 comments

Meta trying to destroy whistleblower Sarah Wynn-Williams, US senator says

07/18 03:31

29 points · 0 comments

Show HN: A zoomable timeline of 4M Wikipedia events

07/18 02:37

I'm building a journal app in Kotlin Multiplatform and for this purpose I have created a zoomable timeline interface.This is a side-project where I reuse the timeline interface to display 4 million events imported from Wikipedia / Wikidata, scored using PageRank. There is more information

Lego building instructions through time

07/18 02:21

4 points · 0 comments

Learning a few things about running SQLite

07/18 01:45

77 points · 19 comments

Homomorphically encrypted CIFAR-10 inference in 200ms

07/18 00:29

24 points · 25 comments

Show HN: Explore the Workspaces of Modern Creators

07/17 23:41

43 points · 37 comments

Frame – Linux X server in Assembly

07/17 23:31

105 points · 64 comments

Short sellers notch $8.7B profit as SpaceX shares dip to IPO price

07/17 23:17

128 points · 97 comments

The state of open source AI

07/17 22:31

305 points · 212 comments

Kimi K3, and what we can still learn from the pelican benchmark

07/17 22:21

194 points · 113 comments

AI Meets Cryptography 2: What AI Found in OpenVM's ZkVM

07/17 22:21

70 points · 0 comments

First atmosphere found on Earth-like planet in habitable zone of distant star

07/17 22:06

277 points · 195 comments

Show HN: Watch bots interact with an SSH honeypot in real time

07/17 22:05

116 points · 44 comments

Three ways people respond to a problem (other than solving it)

07/17 22:00

151 points · 77 comments

A Road to Lisp: Which Lisp

07/17 21:56

144 points · 90 comments

More Bounce to the Ounce

07/17 21:32

86 points · 28 comments

Manufact (YC S25) Is Hiring a Senior infra engineer to build the MCP cloud

07/17 21:22

1 points · 0 comments

VulnHunter: Capital One's agentic AI code security tool

07/17 20:42

49 points · 28 comments

Apple targets dozens of OpenAI employees with legal letters

07/17 20:02

333 points · 280 comments

AWS: Inaccurate Estimated Billing Data – $1.7 billion

07/17 17:42

URL already posted: https://health.aws.amazon.com/health/statusI've got an estimated bill for $1.7 BILLION over this month. Normal usage is < $5.Obvs have created an urgent AWS support ticket. Anyone else seeing something like this?Update: Reddit link: https://w

EEG shows brain can simultaneous encode two speech streams

07/17 13:51

238 points · 155 comments

Pebble Mega Update – July 2026

07/17 11:53

246 points · 160 comments

MoonBASIC: A modern BASIC for building 2D and 3D games

07/15 01:53

14 points · 4 comments

Estimating the heights of New Yorkers from their scuff marks

07/14 07:22

22 points · 5 comments

Latent Space as a New Medium

07/14 02:12

63 points · 18 comments

Frank Lloyd Wright’s first home

07/13 10:23

45 points · 24 comments

Faster binary search: from compiled code to mechanical sympathy

07/12 15:00

51 points · 11 comments

Camera Chase Vehicle

07/10 10:17

177 points · 18 comments

v2.1.212

07/17 08:26

What's changed /fork now copies your conversation into a new background session (its own row in claude agents) while you keep working; the in-session subagent it used to launch is now /subtask Added claude auto-mode reset to restore the default auto-mode configuration, with a confirmation prompt (pa

中文介绍 Anthropic 的 Claude Code 项目发布了 v2.1.212 版本。主要更新包括："/fork" 命令现在会将当前对话复制到一个新的后台会话中，用户可继续工作；原有的会话内子代理功能则被新的 "/subtask" 命令取代。此外，还新增了 "claude auto-mode reset" 功能，用于恢复默认的自动模式配置。

v2.1.211

07/16 07:02

What's changed Added --forward-subagent-text flag and CLAUDE_CODE_FORWARD_SUBAGENT_TEXT environment variable to include subagent text and thinking in stream-json output Fixed permission previews relayed to chat channels not neutralizing bidirectional-override, zero-width, and look-alike quote charac

中文介绍 Anthropic的Claude Code发布v2.1.211版本。更新新增`--forward-subagent-text`标志及`CLAUDE_CODE_FORWARD_SUBAGENT_TEXT`环境变量，用于在stream-json输出中包含子代理文本。同时，修复了聊天频道权限预览中特定字符未能中和的错误。

v2.1.210

07/15 07:45

What's changed Added a live elapsed-time counter to the collapsed tool summary line so long-running tool calls visibly tick instead of looking stuck Added a startup warning for Write(path), NotebookEdit(path), and Glob(path) permission rules — use Edit(path) or Read(path) instead Fixed isolation: 'w

v2.1.209

07/14 14:36

What's changed Fixed /model and other dialogs being blocked in claude agents background sessions (reverts an overly broad guard)

v2.1.208

07/14 09:10

What's changed Added screen reader mode: opt-in plain-text rendering for screen reader users. Run claude --ax-screen-reader, set CLAUDE_AX_SCREEN_READER=1, or add "axScreenReader": true to settings. Added vimInsertModeRemaps setting: map two-key insert-mode sequences like jj to Escape in vim mode Ad

v2.1.207

07/11 08:52

What's changed Auto mode is now available without CLAUDE_CODE_ENABLE_AUTO_MODE opt-in on Bedrock, Vertex AI, and Foundry; disable via disableAutoMode in settings Fixed the terminal freezing and keystrokes lagging while streaming responses containing very long lists, tables, paragraphs, or code block

v2.1.206

07/10 09:45

What's changed Added directory path suggestions to /cd, matching /add-dir behavior Added a /doctor check that proposes trimming checked-in CLAUDE.md files by cutting content Claude could derive from the codebase /commit-push-pr now auto-allows git push to the repo's configured push remote (remote.pu

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

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

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