TypeScript · ★ 16,346 · 🍴 1,473 · 📈 1,057 stars today
TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Skill, LLM-Wiki, Code-Graph) that are governed, shared, and equipped across agents and frameworks.
JavaScript · ★ 82,917 · 🍴 8,891 · 📈 593 stars today
Production-grade engineering skills for AI coding agents.
TypeScript · ★ 4,780 · 🍴 242 · 📈 2,802 stars today
Give your agent a computer 👾
Shell · ★ 206,995 · 🍴 17,878 · 📈 1,873 stars today
Skills for Real Engineers. Straight from my .agents directory.
Python · ★ 23,102 · 🍴 1,773 · 📈 138 stars today
The authentication glue you need.
Python · ★ 2,850 · 🍴 215 · 📈 847 stars today
Lightweight loop engineering state kernel for long-running AI agent teams. Agent-loop agnostic across Codex, Claude Code, and other coding agents, with durable goals, quota-aware auto-wake, executable todos, evidence logs, and verifiable handoffs.
Java · ★ 51,629 · 🍴 11,164 · 📈 13 stars today
Google core libraries for Java
Roff · ★ 77,053 · 🍴 17,407 · 📈 134 stars today
所有小初高、大学PDF教材。
Python · ★ 185,997 · 🍴 46,055 · 📈 37 stars today
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.
Python · ★ 29,009 · 🍴 2,682 · 📈 237 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.
Go · ★ 32,386 · 🍴 2,096 · 📈 888 stars today
DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.
Shell · ★ 268,078 · 🍴 23,963 · 📈 858 stars today
An agentic skills framework & software development methodology that works.
Rust · ★ 12,419 · 🍴 836 · 📈 1,190 stars today
Fast Rust library for PDF inspection, classification, and text extraction. Intelligently detects scanned vs text-based PDFs to enable smart routing decisions.
👍 54
Long-horizon search agents must make multiple sequential actions (steps) to search, retrieve, verify, and integrate evidence to reach a final answer. However, existing methods for training these agents typically treat all steps within a trajectory uniformly during both supervised fine-tuning (SFT) a
中文介绍 针对长时程搜索代理训练中轨迹内所有步骤被统一对待的问题，提出ABSeeker，通过答案回溯的信用分配机制区分各步骤贡献，提升搜索、检索、验证与整合证据的多步决策能力。
👍 45
Text-to-image (T2I) models can produce visually compelling images, yet they remain limited on open-world tasks that require complex semantic understanding, multi-step reasoning, and the integration of external world knowledge. Existing efforts introduce agent capabilities into image generation, but
中文介绍 针对文生图模型在开放世界任务中缺乏复杂语义理解、多步推理和外部知识整合的问题，提出ToolArtist，将工具使用能力引入统一多模态模型，实现代理式图像生成，通过调用外部工具增强生成质量与任务完成能力，弥补现有方法的不足。
👍 37
Personalized LLMs with persistent memory are increasingly deployed, yet the faithfulness of their user models remains unexamined. We study over-inference (OI): the phenomenon where LLMs fabricate user attributes beyond what evidence supports. We introduce MirageBench, comprising 150 personas balance
中文介绍 研究个性化LLM的过度推断现象：模型会捏造超出证据支持的用户属性。引入MirageBench基准，包含150个平衡人格，揭示自监控机制在识别此类捏造时可能产生误导，并对记忆增强个性化系统的可靠性提出警示。
👍 28
LLM agents are increasingly applied to open-ended everyday requests that span work, study, and life. These tasks are long-horizon, cross-environment, and multimodal, forcing the agent to preserve goals and constraints across many steps while navigating heterogeneous tools and attachments. While prio
中文介绍 面向跨环境、多模态的长期日常任务，提出OneDayAgent，一种长时程自治代理框架，帮助LLM代理在多步交互中保持目标与约束，并处理异构工具和附件，推进自主代理在真实开放场景中的应用。
👍 21
Agent self-evolution updates an agent's persistent state from prior experience and reuses it to solve related tasks more effectively. Evaluating self-evolution is difficult: existing benchmarks provide limited coverage of economically valuable task domains, do not always design training and test tas
中文介绍 提出GDPevo，用于评估代理自我进化在真实商业任务上的表现。现有基准在经济价值任务覆盖和训练/测试任务设计上存在不足，GDPevo通过更贴合真实业务的设置衡量代理从经验中更新状态并复用知识的能力。
👍 20
Long-horizon reasoning in recent LLMs demands that the model switch between distinct skills inside a reasoning chain, such as first doing a math derivation, then using the result to plan a schedule. We call such problems cross-skill long-horizon tasks: multi-step tasks whose steps require different
中文介绍 提出技能熵度量，用于基准测试和训练跨技能长时程推理。这类任务要求模型在推理链中切换不同技能，如先数学推导再制定计划，该指标可衡量技能切换复杂度并指导训练，为技能原生LLM的开发提供新视角。
👍 19
On-Policy distillation (OPD) transfers teacher capabilities by supervising student-sampled trajectories with dense token-level teacher signals. Recent selective OPD methods improve this process by prioritizing signals that are confident, informative, or learnable. However, the assumptions overlook a
中文介绍 针对在线策略蒸馏中教师信号可能包含虚假相关而误导学生的问题，提出虚假信号感知的OPD方法。现有选择性蒸馏只关注置信度、信息量或可学习性，忽略了信号相关性，新方法通过识别并去伪提升蒸馏效果。
👍 15
While instruction-based video editing has advanced rapidly, real-world videos contain tightly coupled audio and visual signals, and editing one modality often requires coordinated changes in the other. Existing benchmarks primarily evaluate visual transformations on silent clips or isolated audio ed
中文介绍 提出AVE-Compass，面向音视频联合编辑能力的整体评估基准。现有基准多针对静音视频或孤立音频，忽略真实视频中音视频信号的强耦合；AVE-Compass衡量指令引导下跨模态协调编辑能力，推动更全面的音视频编辑评测。
👍 14
Video world models predict future observations conditioned on historical observations and control signals, enabling long-horizon generation through autoregressive state transitions. Unlike conventional video generation models that primarily capture visual appearance and motion, video world models le
中文介绍 提出MiniWorld，降低从零训练视频世界模型的成本，使其更普及。视频世界模型根据历史观测与控制信号预测未来，支持自回归状态转换的长期生成。MiniWorld在保证建模能力的同时简化训练流程，赋能更多研究者。
👍 13
This technical report presents K-EXAONE 2.0, an open-weight multilingual foundation model developed by LG AI Research as a step in our effort toward global frontier-scale foundation models. Rather than training from scratch, we upcycle K-EXAONE and expand its architecture, yielding a Mixture-of-Expe
中文介绍 LG AI Research发布K-EXAONE 2.0，一个开源权重的多语言基础模型。采用升级循环方式而非从零训练，扩展架构得到Mixture-of-Experts模型，目标是迈向全球前沿规模基础模型。
👍 13
We introduce a new problem domain for human action recognition: the fine-grained analysis of children's gait behaviors from standard RGB video. We specifically target the ambulatory patterns of children aged 3-17 years. Such behaviors arise naturally in the diagnosis and treatment of several critica
中文介绍 提出人类动作识别新任务：从标准RGB视频中对3-17岁儿童步态行为进行细粒度分析。该任务可服务于多种疾病的诊断与治疗，为医疗场景下的儿童步态识别提供新的研究方向。
👍 12
Reinforcement learning with verifiable rewards (RLVR) has become a standard paradigm for post-training large language models (LLMs). While Group Relative Policy Optimization (GRPO) is widely adopted, it suffers from sparse reward signals and loses gradients entirely when all responses within a group
中文介绍 针对GRPO在组内所有响应均错误时梯度完全消失、奖励信号稀疏的问题，提出「在失败处蒸馏」方法，利用自适应教师引导恢复负RL组的可学习信号，改善可验证奖励强化学习（RLVR）中LLM的后训练效果。
👍 12
Memory-augmented VLM agents act on persistent spatial knowledge, yet that knowledge silently goes stale as the environment changes. We ask what happens when an agent must reconcile a confident memory claim with a contradicting observation, and whether current models can catch the conflict before it
中文介绍 实证研究记忆增强的视觉语言模型代理中空间记忆过时问题：当环境变化后，代理需要调和自信记忆与矛盾观察，考察现有模型能否在冲突导致失败前察觉，揭示记忆不可靠时的行为模式与局限。
👍 11
Do Large Language Models (LLMs) possess genuine structural reasoning, or merely rely on surface-level pattern matching? The financial domain, demanding numerical precision and multi-step logic over long contexts, is an ideal testbed. Existing benchmarks fail to capture real-world industrial complexi
中文介绍 检验LLM在金融领域长时程陈述上是否具备真正的结构化推理，而非表面模式匹配。现有基准未覆盖真实工业复杂度，本文设计贴近实际的长上下文金融推理测试，评估模型数值精度与多步逻辑能力。
👍 11
Modern agent frameworks equip large language models with external skill libraries to solve complex tasks. However, it remains unclear whether these systems can effectively evolve their skills and whether the resulting skills improve task-solving capabilities. To bridge this gap, we introduce Continu
中文介绍 提出ContinualSkillBench，评估LLM代理能否真正进化技能库并提升任务解决能力。现代代理框架引入外部技能库，但技能进化机制及其效果尚不明确，该基准填补了这一评估空白。
👍 10
GUI agents must remember both useful experience from earlier tasks and unfinished progress in the current interaction. Latent memory offers a compact solution by compressing multimodal trajectories into a few continuous tokens. Existing methods, however, usually map each trajectory to one fixed memo
中文介绍 面向GUI代理的潜在记忆，提出FocusMem，将内容、读出与信任三个因素解耦。现有方法通常将每个轨迹映射为固定记忆，FocusMem通过分解设计更灵活地压缩多模态轨迹，兼顾历史经验与当前交互进度。
👍 10
Query-agnostic KV cache eviction compresses a context once and reuses the resulting cache for arbitrary future queries, but performance can collapse under tight budgets. Existing methods primarily improve which original KV pairs are retained. We introduce RestoreKV, which complements this selection-
中文介绍 提出RestoreKV，用于在激进的查询无关KV缓存驱逐下恢复完整缓存行为。现有方法仅改进保留哪些原始KV对；RestoreKV补充选择策略，通过补偿被驱逐信息减轻性能坍塌，在紧预算下提升缓存压缩质量。
👍 9
Skill-based prompting has become a practical mechanism for improving large language model (LLM) agents, yet existing skill acquisition methods often treat skills as experience summaries, memory entries, or direct summaries of successful demonstrations. This creates a mismatch for weaker student agen
中文介绍 提出SKILL-KD，一种用于LLM代理的对比技能蒸馏方法。现有技能获取方式常将技能视为经验摘要或成功示范的直接概括，导致弱学生代理不匹配；SKILL-KD通过对比学习从教师模型中提炼可迁移技能，提升学生代理的技能应用能力。
👍 8
Model checkpoints are growing in both number and size, which makes archival, transfer, and deployment increasingly costly. General-purpose compressors can reduce storage requirements but ignore tensor structure, whereas existing tensor-specific compressors rely on fixed and format-specific pipelines
中文介绍 将无损张量压缩形式化为程序合成问题，提出新方法：通过生成专门针对张量结构的压缩程序，兼顾通用压缩器的灵活性与张量压缩器的结构性，降低模型检查点的存储与传输成本。
👍 8
On-Policy Self-Distillation (OPSD) has become a standard post-training approach for improving visual reasoning in multimodal large language models (MLLMs). Existing methods draw privileged information from diverse input sources to guide self-distillation. Yet these designs overlook Modality Imbalanc
中文介绍 提出OPD-V，一种带模态平衡的视觉在线策略自蒸馏方法。现有OPSD方法从多种输入源提取特权信息，但忽略模态不平衡问题；OPD-V通过平衡多模态信号，改善多模态大语言模型的视觉推理后训练效果。
@XFreeze · 251.7K 粉丝 · 778.3K 阅 · 558 赞 · 80 转
Grok Build: What It’s Actually For Most people underestimate it by an order of magnitude. Ask someone what an AI coding agent is for, and they’ll probably tell you it builds apps. That’s true and it
中文介绍 Grok Build 的真正价值被多数人低估了一个数量级——它不只是写应用的 AI 编程代理，更可被当作你的「自主员工」。文章给出将其用于实际工作流、替代重复任务的方法。
@tetsuoai · 236.6K 粉丝 · 493.9K 阅 · 851 赞 · 138 转
Grok Imagine crossed a line this week. Imagine Video 1.5 shipped last month with better motion, physics, and audio. On July 31 xAI added the piece that makes actual filmmaking possible: image and
中文介绍 以 Grok Imagine 从构思到成片制作短片的完整流程。Imagine Video 1.5 已显著改进运动、物理和音频表现，xAI 又在 7 月 31 日加入图像生成能力，补齐了真正电影制作的关键一环：先出分镜/概念图，再生成动态视频。
@sawyerhood · 17.1K 粉丝 · 185.1K 阅 · 519 赞 · 27 转
I'm excited to show something I've been working on recently: bb, an agentic IDE that builds itself. This started as a passion project by @_ymichael, and over time I changed from being an early user
中文介绍 展示 bb——一个会自我构建的 agentic IDE。项目最初是 @_ymichael 的业余作品，作者从早期用户一路深度参与，最终参与开发。核心想法是让 IDE 借助 agent 能力自动生成代码、重构自身，重新定义开发工具。
@RaoulGMI · 1.5M 粉丝 · 90.0K 阅 · 530 赞 · 72 转
Within about two years, most economic activity on Earth won't have a human anywhere in it. Think about a single trade today. Someone decides to make it, another person approves it, someone books it,
中文介绍 预判约两年内，地球上大部分经济活动将不再需要人类参与：从发起交易、审批到记账，整条链路都由 AI 完成。文章称之为「隐形经济」，并讨论人类被移除后金融与商业的深层影响。
@VibeMarketer_ · 29.9K 粉丝 · 56.6K 阅 · 501 赞 · 30 转
I'm going to show you how to automate your content ideation system using Hermes kanban, a new multi-agent feature from Nous Research. Every night, it will monitor a mix of 50+ X accounts, newsletters,
中文介绍 用 Nous Research 新推出的多智能体功能 Hermes kanban 搭建自主内容引擎：每晚自动监控 50+ X 账号和 newsletter，筛选趋势并批量生成选题，形成从信息收集、过滤到排期的无人化内容流水线。
@YifeiChen1121 · 224 粉丝 · 51.8K 阅 · 567 赞 · 22 转
How nature and serif fonts became the default face of artificial intelligence TLDR: AI branding converged on serifs + nature photos + soft neutrals to feel "humane." It worked… until everything
中文介绍 拆解 AI 品牌为何普遍选择衬线字体、自然照片和柔和中性色：这套视觉语言为营造「人性化」而高度趋同，结果反而成为 AI 内容的标准脸谱，连看似有品位的 AI slop 也难以逃脱同质化。
@yisongyue · 24.2K 粉丝 · 48.3K 阅 · 503 赞 · 37 转
A new scaling dimension is emerging Today, we scale models through better data, architectures, and compute. We scale agents through better tools, search, verification, and harnesses. Tomorrow, we will
中文介绍 提出「知识飞轮」将成为新的扩展维度：当前用数据、架构和算力扩展模型，用工具、搜索、验证和 harness 扩展 agent；下一步则是通过知识积累、复用与迭代获得规模化收益。
中文介绍 谷歌DeepMind推出气象AI模型WeatherNext，称其在气旋预报方面取得突破性进展。该模型有望提升热带气旋等极端天气的预测能力与时效，为防灾和气象预报提供更强工具。
ChatGPT introduces improved GPT-5.6 Sol with better accuracy and consistency, plus expanded access for free users and unlimited everyday chats with GPT-5.6 Luna.
中文介绍 OpenAI在ChatGPT中推出改进版GPT-5.6 Sol，提升准确性和一致性；同时向免费用户扩大GPT-5.6 Luna的使用权限，并开放无限次日常聊天。
OpenAI and the American Psychological Association advance evidence-based guidance, resources, and safeguards for responsible AI use and youth mental health.
中文介绍 OpenAI与美国心理学会（APA）合作，围绕青少年心理健康和AI应用，共同推进基于证据的指导建议、资源与保障措施，确保AI被负责任地使用。
The end of an era.
中文介绍 据Latent Space报道，Jeff、Sanjay、Oriol和Quoc离开Google DeepMind；Demis Hassabis将出任董事会主席，Koray Kavukcuoglu担任高级副总裁，领导层变动引发关注。
New OpenAI Signals data shows how people use ChatGPT worldwide, with country-level insights on adoption, usage trends, and evolving behavior.
中文介绍 OpenAI发布新的Signals数据，展示全球用户使用ChatGPT的实际情况，包括各国采用率、使用趋势以及用户行为演变，并提供国家层面的洞察。
中文介绍 Hugging Face博客发布文章，介绍Baseten接入其推理服务提供商（Inference Providers）平台，用户可通过该平台使用Baseten部署的模型服务。
中文介绍 本期TLDR AI简报涵盖：谷歌DeepMind管理层调整，Meta推出Muse Code编程工具，Anthropic组建芯片团队，以及相关人工智能行业要闻。
**Google DeepMind** undergoes a leadership reshuffle with **Demis Hassabis** moving to Chair and Chief Scientist roles, while **Koray Kavukcuoglu** takes operational control focusing on **Gemini** and product execution. The launch of **Discovery Loop** by founders including **Jeff Dean**, **Sanjay G
中文介绍 Google DeepMind进行领导层重组：Demis Hassabis转任董事长兼首席科学家，Koray Kavukcuoglu接管运营，聚焦Gemini与产品执行；同时创始团队推出Discovery Loop。
A quiet day lets us highlight a Cursor launch and an engineering debate
中文介绍 Latent Space发文讨论“Megakernels已死又复活”的工程话题，并在较为平静的一天里重点关注Cursor的发布和一场关于内核编程的工程辩论。
中文介绍 本期TLDR AI简报涵盖：谷歌推出LLM路由器，Cloudflare推出钱包功能，以及Anthropic与Volta达成合作等人工智能领域要闻。
OpenAI explains recent third-party cybersecurity evaluation incidents and outlines new safeguards to strengthen AI model testing and evaluation.
中文介绍 OpenAI就近期涉及自身模型的第三方网络安全评估事件进行说明，并公布新的防护措施，以加强AI模型测试与评估环节的安全性。
An external reconstruction of how Memory, Proactivity, Scheduling, Browser Use, Plugins, Skills and Tools work in the new ChatGPT Work.
中文介绍 Latent Space发文解析ChatGPT Work，从外部重构其记忆、主动性、日程安排、浏览器使用、插件、技能和工具等核心功能，称其为面向十亿用户的智能体。
中文介绍 Liquid AI发布LFM2.5-2.6B模型，目标是让AI代理能够随处部署在本地设备上，降低对云端算力的依赖，为终端应用提供更灵活的推理方案。
**Alibaba** launched **Qwen3.8-Max**, enhancing multimodal capabilities and agent ecosystem integration. **NVIDIA** introduced **Alpamayo 2 Super** for autonomous vehicle reasoning, while **Mistral AI** released **Shieldstral**, a 3B parameter open-weights safety model for on-device moderation. **Po
中文介绍 阿里发布Qwen3.8-Max，增强多模态能力与Agent生态集成；NVIDIA推出Alpamayo 2 Super用于自动驾驶推理；Mistral AI发布3B参数开源安全模型Shieldstral。
Qwen is so back!
中文介绍 阿里发布Qwen 3.8 Max（2.4T参数）和27B等新开源权重模型，重点面向编程与协作场景，引发社区广泛关注。
美股大盘整体偏强：SPY 距 52 周高点仅 0.42%，MACD 金叉且多头排列，NVDA 5 日涨 12.28% 动能突出；但结构分化，MSFT RSI 78.1 超买、TSLA 空头排列且 MACD 死叉、META 距高点回撤 25.92%，显示资金集中在少数权重股，QQQ RSI 54.3 中性震荡。加密市场受恐慌贪婪指数 29 影响情绪谨慎，总市值 2.28 万亿美元、24 小时微跌 0.61%，BTC 主导率 56.6%；BTC 站上 SMA20/50 但受制于 SMA200，ETH RSI 54.8 相对抗跌，SOL 空头排列偏弱。中概板块 5 日集体走强，BABA 涨 9.02%、PDD 涨 3.9%，JD RSI 73.4 超买，腾讯 MACD 金叉。商品外汇方面，黄金期货 1 日涨 1.33% 且 MACD 柱转正，WTI 原油 1 日大涨 3.99% 但 5 日仍跌 6.42% 且 MACD 死叉，美元/人民币逼近 52 周低点、RSI 33.2 偏弱。
MSFT 现价 499.86 美元，RSI 14 为 78.1 进入超买区，5 日涨幅 10.81%、1 日涨幅 2.54%，动能强劲。MACD 柱 25.45 高于信号线 13.44，多头动能持续发散。价格远高于 SMA20（418.09）与 SMA50（405.27），呈明显多头排列。需注意股价距 52 周高点仍低 9.73%，且超买状态下短线回调压力可能积聚。
京东现价 32.81 美元，RSI 14 为 73.4 进入超买区，MACD 金叉且柱值 1.25 高于信号线 1.05。价格高于 SMA20（30.88）、SMA50（28.97）与 SMA200（29.44），均线多头排列。5 日涨幅 1.55%，距 52 周高点仅低 10.99%，技术面偏强。但 RSI 超买提示短线追高风险，动量可能暂时透支。
特斯拉现价 319.53 美元，RSI 14 为 36.9 处于偏弱区域，MACD 柱 -21.39 低于信号线 -21.04，死叉状态延续。价格低于 SMA20（347.29）、SMA50（382.68）与 SMA200（409.39），呈空头排列。1 日跌 0.63%，距 52 周高点回撤 35.94%，技术结构整体偏弱，中期趋势承压。
Solana 现价 72.56 美元，RSI 14 为 42.5 偏弱，MACD 柱 -0.80 低于信号线 -0.65，维持死叉。价格低于 SMA20（74.71）、SMA50（75.04）与 SMA200（84.34），空头排列明确。1 日跌 1.89% 为主流加密中最大跌幅，距 52 周高点低 71.34%，技术面呈下行趋势。
比特币现价 64,219 美元，RSI 14 为 51.1 处于中性区间，1 日微跌 0.59%、5 日涨 2.32%。价格高于 SMA20（64,387）与 SMA50（63,238），但低于 SMA200（70,508），中期趋势尚未扭转。MACD 柱 -5.04 低于信号线 22.16，短期动能偏弱。恐慌贪婪指数 29 处于恐慌区，整体方向不明。
VIX 恐慌指数
10Y 美债收益率 (%)
美元指数 DXY
S&P 500 ETF
Nasdaq 100 ETF
Apple
Microsoft
Nvidia
Alphabet
Tesla
Meta
Bitcoin
Ethereum
Solana
阿里巴巴 (BABA)
拼多多 (PDD)
京东 (JD)
腾讯控股 (0700.HK)
黄金期货
WTI 原油期货
美元 / 人民币
本报告基于公开行情数据计算的技术指标，仅供技术指标解读参考，不构成任何投资建议。过去走势不代表未来表现，技术指标存在滞后性与失效可能，市场受宏观事件、流动性等多重因素影响。请读者结合自身情况独立判断，据此操作风险自担。
The two men, users of the platform Kick, received suspended sentences and fines.
中文摘要 法国一家法院对两名Kick直播平台用户作出有罪判决，两人获缓刑并处罚款，案件涉及该平台上一场直播死亡事件。具体刑罚细节未公布。
Deputy prime minister says Burke’s speech before National Press Club was not cancelled. Follow today’s news live Federal parliament is returning from winter break next week. Do you have a question you’d like to ask Guardian Australia’s team of political reporters in Canberra? Our political editor, T
中文摘要 澳大利亚副总理理查德·马尔斯坚称，托尼·伯克关于工党新限制性移民政策的演讲并未被取消，该演讲原计划在国家记者俱乐部举行。
A New Mexico judge ordered the Instagram, Facebook and Whatsapp parent company to pay another $567m for the way it has harmed children.
中文摘要 美国新墨西哥州一名法官裁定，Meta（Instagram、Facebook和WhatsApp母公司）因在儿童保护方面造成伤害，需额外支付5.67亿美元罚款，这是针对该社交媒体巨头的最大儿童安全裁决。
The US, Iran, and Oman are close to a temporary deal to reopen the Strait of Hormuz
中文摘要 美国总统特朗普表示战争不能再“持续太久”，并透露美国、伊朗和阿曼已接近达成一项临时协议，以重新开放霍尔木兹海峡。
A review of the body formerly charged with reducing government spending has found several issues with its calculations.
中文摘要 美国监管机构审查发现，政府效率部（DOGE）报告的1100亿美元节省中，部分数据有误或缺乏证据。该机构此前负责削减政府开支，审查指出其计算存在多项问题。
Blockades and bureaucracy turn education into an uphill journey for talented Gaza students seeking a brighter future.
中文摘要 由于来自以色列和欧洲的旅行限制，加沙学生被困当地。封锁和繁琐手续让许多才华横溢的加沙学生为追求更光明未来的求学之路变得异常艰难。
People in South Lebanon question their government's approach as Israel launches more attacks and displacement orders.
中文摘要 随着以色列恢复对黎巴嫩南部的打击并发布更多撤离令，当地居民感到愤怒，质疑本国政府的应对方式，直呼“没有红线”。
US President Donald Trump signed two executive orders to further limit who is eligible for birthright citizenship.
中文摘要 美国总统特朗普签署两项行政命令，进一步限制有资格获得出生公民权的人群，禁止所谓的“生育旅游”。此举将影响赴美生子人群。
The Iranian-backed militia said it had targeted the armed forces in two resource-rich provinces, as escalating violence in Yemen imperils a fragile truce.
中文摘要 伊朗支持的胡塞武装声称袭击了也门两个资源丰富省份的政府军。随着暴力升级，也门脆弱的停火面临威胁，内战或重新爆发。
US president says he is involved in talks and war will end 'pretty soon', claiming that Iran cannot 'go much longer'.
中文摘要 伊朗议会议长卡利巴夫指责特朗普循环上演“表演式外交”。特朗普声称正参与谈判，战争将“很快”结束，并称伊朗无法“再撑太久”。
Michigan progressive fended off AIPAC as well as Democratic establishment in US Senate race, supporters say.
中文摘要 支持者称，密歇根州进步派人物阿卜杜勒·埃尔-赛义德在美国参议院选举中战胜了美国以色列公共事务委员会（AIPAC）和民主党建制派，这场“非凡”胜利震惊了亲以色列建制派。
Canada’s prime minister laughed off Trump’s description of Canada but suggested that trade talks between the two countries are acrimonious.
中文摘要 美国总统特朗普称加拿大人“讨厌”后，加拿大总理卡尼一笑置之，但表示这个词用来形容两国贸易谈判倒也贴切，暗示谈判气氛不睦。
The Houthis say they launched missiles and drones at "Saudi military build-ups" in the central provinces of Marib and Hadramawt.
中文摘要 据报道，胡塞武装袭击造成至少30名也门政府军死亡。胡塞方面称，他们向马里卜省和哈德拉毛省中部的“沙特军事集结”发射了导弹和无人机。
India's top court has put the spotlight on unsafe footpaths and the country's forgotten pedestrians.
中文摘要 印度最高法院聚焦不安全的步行道和该国被遗忘的行人问题。报道指出，印度本是一个步行者众多的国家，但其城市却忽视了行人需求。
Al Jazeera's Nils Adler reports on how far-right 'Active Clubs' use fitness and camaraderie to recruit young white men.
中文摘要 专家警告，极右翼“活跃俱乐部”以健身和情谊为幌子，招募年轻白人男性从事极端主义活动。半岛电视台记者尼尔斯·阿德勒就此进行了报道。
Iran will seek to bar US and Israeli ships from the Strait of Hormuz and require compensation from hostile countries before they’re allowed to use it, according to local media reports on a proposed Iran-Oman deal. Bloomberg's Michael Heath reports. (Source: Bloomberg)
中文摘要 据当地媒体报道，伊朗拟在与阿曼的协议中寻求禁止美国和以色列船只进入霍尔木兹海峡，并要求敌对国在使用前支付赔偿。这是相关谈判中的一项新动向。
Mexican authorities detained a former governor over charges he concealed evidence in the more than decade-old abduction and presumed murder of dozens of students training to be teachers, one of the country’s most infamous criminal cases.
中文摘要 墨西哥当局逮捕一名前州长，指控他在十多年前师范学生遭绑架及推定谋杀案中隐瞒证据。该案是墨西哥最臭名昭著的刑事案件之一。
A New Mexico judge ordered the Instagram, Facebook and Whatsapp parent company to pay another $567m for the way it has harmed children.
中文摘要 美国新墨西哥州法官裁定，Meta（Instagram、Facebook和WhatsApp母公司）因伤害儿童需再支付5.67亿美元罚款，这是针对社交媒体公司最大规模的儿童安全裁决。
The yen is heading into the end of the week having surrendered nearly half of its intervention-driven gains, fueling speculation among traders that authorities may step into the market again.
中文摘要 本周日元兑美元回吐了此前美日联合干预所带来涨幅的近一半，交易员因此猜测当局可能再次入场干预，汇率或面临进一步波动。
DBS Group CEO Tan Su Shan sees AI costs trending down with more use of tokens, driven by what she calls the 'paradox of token spend.' She emphasizes that the bank maintains an open architecture, staying open-minded and avoiding being 'too wedded' to a model or any one provider. In a wide-ranging int
中文摘要 新加坡星展集团首席执行官陈淑珊表示，随着代币使用增加，人工智能成本将呈下降趋势，她称之为“代币支出悖论”。她强调该行保持开放架构，避免过度依赖单一模型或供应商。
Gold moved in a narrow range as traders assessed the impact of a reported Iranian attack in the Strait of Hormuz on the Federal Reserve’s path for interest rates.
中文摘要 现货黄金在窄幅区间内交投，表现平稳。交易员正评估据报伊朗在霍尔木兹海峡发动袭击对美联储利率路径的影响，以判断金价后续方向。
Federal Reserve Bank of St. Louis President Alberto Musalem explains why he expressed a preference to raise rates at the Federal Open Market Committee's July meeting. Speaking at an event in São Paulo, Musalem also discusses the outlook for inflation. (Source: Bloomberg)
中文摘要 美国圣路易斯联储行长阿尔贝托·穆萨勒姆在圣保罗的活动上表示，他在7月联邦公开市场委员会会议上倾向于加息，并解释了原因，同时讨论了通胀前景。
The airline industry wants passengers to be careful when travelling with lithium ion batteries.
中文摘要 据英国广播公司商业新闻，航空公司正就锂离子电池发出安全警告，提醒乘客在乘坐飞机旅行时谨慎携带此类电池，注意遵守相关安全规定，以降低飞行过程中的潜在风险。
The retailer has been around for 50 years but struggled to compete with newer rivals such as Amazon. Can its new owners turn it around?
中文摘要 英国零售品牌Argos已成立50年，一直难以与亚马逊等较新竞争对手抗衡。品牌正在进行改造，其新东家能否扭转局面、吸引新顾客仍待观察。
Federal Reserve Bank of St. Louis President Alberto Musalem says policymakers can't afford to tolerate higher inflation while they wait for the possibility of stronger productivity growth. He speaks at an event in São Paulo. (Source: Bloomberg)
中文摘要 圣路易斯联储行长穆萨勒姆在圣保罗表示，政策制定者不能在等待生产率强劲增长可能性的同时容忍更高通胀，认为需要对通胀施加“有意义的约束”。
China is accelerating its push to internationalize the yuan as domestic banks add new foreign currencies available for direct settlement to bypass the dollar in cross-border trade.
中文摘要 中国正加快推动人民币国际化，国内银行扩大可直接结算的外币种类，以在跨境贸易中绕开美元，助力提升人民币的全球作用。
The ecommerce and cloud company hopes to recoup some of its lost shine with Monday’s release
10 回复 · 程序员 节点
11 回复 · 程序员 节点
7 回复 · 程序员 节点
11 回复 · 程序员 节点
25 回复 · Apple 节点
18 回复 · Apple 节点
13 回复 · Apple 节点
15 回复 · Apple 节点
11 回复 · Linux 节点
31 回复 · Apple 节点
该源今日无内容。