JavaScript · ★ 31,079 · 🍴 1,935 · 📈 3,902 stars today
Agent skill for beautiful, verifiable architecture, workflow, sequence, data-flow, and lifecycle diagrams—self-contained HTML with motion and crisp export.
中文介绍 Archify 为 AI agent 提供绘制架构图、流程图、时序图、数据流图和生命周期图的技能，输出自带动画、可验证且可干净导出的独立 HTML 图表，适合需要生成可视化文档的开发场景。
JavaScript · ★ 12,635 · 🍴 2,502 · 📈 1,855 stars today
A spy satellite simulator in your browser, except the data is real. Live open source spatial intelligence on a photorealistic 3D globe.
中文介绍 在浏览器中模拟间谍卫星，但使用真实数据。基于照片级逼真的 3D 地球展示实时开源空间情报，可用于地理空间数据可视化、目标监测与态势感知，适合研究人员和地图爱好者。
Python · ★ 37,950 · 🍴 3,574 · 📈 1,587 stars today
Turn any AI agent into an AI Scientist. The #1 Agent Skills library for science, used by 190,000+ scientists worldwide. 165 ready-to-use validated skills plus 100+ scientific databases covering biology, chemistry, medicine, and drug discovery. Compatible with Cursor, Claude Code, Codex, Pi, Antigrav
中文介绍 将任意 AI agent 变成 AI 科学家的技能库，包含 165 个经验证的技能和 100+ 科学数据库，覆盖生物学等领域，已被全球 19 万科学家使用。适合科研人员快速调用专业工具。
Go · ★ 3,493 · 🍴 104 · 📈 789 stars today
like netcat, but over Tailscale's data plane, without Tailscale's control plane
中文介绍 类似 netcat 的网络工具，但通过 Tailscale 数据面传输，不依赖 Tailscale 控制面。适合在 Tailscale 网络内快速建立原始 TCP/UDP 连接，用于调试、端口转发和网络探测。
TypeScript · ★ 22,250 · 🍴 4,309 · 📈 907 stars today
Open Multi-Agent Interactive Classroom — Get an immersive, multi-agent learning experience in just one click
中文介绍 清华开源的多智能体交互课堂，一键获得沉浸式多智能体学习体验。用于构建 AI 助教、多角色讨论或模拟交互环境，适用于教学和协作式学习场景。
Python · ★ 28,699 · 🍴 3,163 · 📈 150 stars today
Fully automatic censorship removal for language models
中文介绍 全自动移除语言模型审查限制的工具，无需人工干预即可规避模型输出中的过滤规则。主要用于 AI 安全研究、模型行为分析和审查机制测试。
JavaScript · ★ 49,119 · 🍴 1,640 · 📈 32 stars today
</> htmx - high power tools for HTML
中文介绍 htmx 是 HTML 增强工具，通过在属性中直接声明 AJAX、SSE、WebSocket 等交互能力，让后端返回的 HTML 片段动态更新页面，无需复杂 JavaScript。适合追求简洁的服务端渲染架构。
Go · ★ 2,871 · 🍴 81 · 📈 303 stars today
Help AI coding agents write modern Go
中文介绍 JetBrains 出品的现代 Go 编码指南，帮助 AI 编码代理写出符合当前最佳实践的 Go 代码。涵盖项目结构、并发、错误处理等约定，适合 Go 开发者及使用 AI 辅助编程的团队。
Python · ★ 73,922 · 🍴 8,470 · 📈 73 stars today
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows
中文介绍 精选的 Claude Skills 资源与工具清单，用于定制 Claude AI 工作流。收录各类技能、示例和最佳实践，帮助开发者快速扩展 Claude 能力并构建自动化任务。
Python · ★ 54,068 · 🍴 6,714 · 📈 806 stars today
World's first open-source, agentic video production system. 12 production pipelines, 100+ tools, 700+ agent skill and production-knowledge files. Turn your AI coding assistant into a full video production studio.
中文介绍 开源智能体视频制作系统，提供 12 条制作流程、100+ 工具和 700+ 技能文件，可将 AI 编码助手变为完整视频制作工具。适合自动化剪辑、脚本生成等内容生产场景。
TypeScript · ★ 14,646 · 🍴 1,756 · 📈 517 stars today
Open source alternative to Semrush and Ahrefs
中文介绍 开源 SEO 分析工具，作为 Semrush 和 Ahrefs 的替代方案。提供关键词研究、反向链接分析、排名追踪等功能，适合站长和 SEO 从业者自建数据平台。
Python · ★ 4,918 · 🍴 741 · 📈 35 stars today
Turn your PC, Mac, or Linux box into an AI server. LLM inference, chat UI, voice, agents, workflows, RAG, and image generation.
中文介绍 将 PC、Mac 或 Linux 设备变为 AI 服务器，支持 LLM 推理、聊天界面、语音、Agent、工作流、RAG 和图像生成。适合本地部署和私有化 AI 服务。
Python · ★ 76,039 · 🍴 9,261 · 📈 550 stars today
Drop in a screenshot and convert it to clean code (HTML/Tailwind/React/Vue)
中文介绍 将截图自动转换为干净的代码，支持 HTML/Tailwind/React/Vue。通过视觉识别还原页面结构和样式，适合前端开发者快速实现设计稿或灵感原型。
Python · ★ 35,407 · 🍴 3,958 · 📈 358 stars today
Official, Anthropic-managed directory of high quality Claude Code Plugins.
中文介绍 Anthropic 官方维护的高质量 Claude Code 插件目录，集中收录经过审核的插件，方便开发者发现、安装和管理 Claude Code 扩展，提升 AI 编程工作流效率。
C++ · ★ 39,320 · 🍴 10,878 · 📈 128 stars today
GoogleTest - Google Testing and Mocking Framework
中文介绍 GoogleTest 是 Google 开源的 C++ 测试与模拟框架，支持单元测试、断言、测试夹具和 mock 对象，广泛用于 C++ 项目的自动化测试。适合需要可靠测试体系的 C++ 开发者。
TypeScript · ★ 8,803 · 🍴 2,756 · 📈 5 stars today
Action for checking out a repo
中文介绍 GitHub Actions 官方 checkout 操作，用于在 workflow 中检出仓库代码，支持分支、标签、子模块、SSH 等参数。是几乎所有 CI/CD 工作流的基础步骤，便于后续构建、测试和部署。
Python · ★ 3,557 · 🍴 409 · 📈 39 stars today
🕵️♂️ (2-in-1) Email & Username OSINT suite for deep data extraction just from a single Email/Username. Analyzes 455+ actively maintained scan vectors (175+ email / 280+ username) for security research, investigations, and digital footprinting.
中文介绍 开源 OSINT 套件，可仅凭邮箱或用户名进行深度数据挖掘，分析 455+ 活跃扫描向量（175+ 邮箱 / 280+ 用户名），用于安全研究和信息收集。适合渗透测试与威胁情报分析。
JavaScript · ★ 90,712 · 🍴 9,710 · 📈 196 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 面向 AI 编码代理的生产级工程技能集合，提供可靠的代码生成、测试、重构等工程实践，帮助 AI 代理在真实项目中输出高质量代码。适合在 CI/CD 或开发流程中集成 AI 辅助。
Go · ★ 2,716 · 🍴 77 · 📈 284 stars today
Model router for agentic systems. Routes every prompt to the right model in <50ms. Cut costs 40-70% with just an endpoint change.
中文介绍 面向智能体系统的模型路由器，能在 50ms 内将每个提示路由到合适模型，仅需修改端点即可降低 40-70% 成本。适合管理多模型调用的 agent 应用，优化性能和费用。
👍 170
Recent studies suggest that video generation models can exhibit certain forms of zero-shot visual reasoning through generated frames. Yet reliable evaluation remains challenging: benchmarks should adopt inputs aligned with the visual priors of current video models, require valid evolving processes r
中文介绍 提出VGI-Bench基准，评估视频生成模型的零样本视觉推理能力。针对现有评测难以对齐模型视觉先验、缺乏有效演化过程的问题，设计新的评测方法，为视频生成模型的视觉智能提供更可靠的量化衡量。
👍 135
Massively parallel simulation changes the data regime in which off-policy reinforcement learning (RL) is trained, challenging stabilizers designed for data-limited replay. Through controlled experiments across eight benchmark families, we show that these stabilizers are data-regime-dependent: parame
中文介绍 研究大规模并行模拟下离策略强化学习的训练稳定性。通过八个基准族的对照实验发现，传统稳定器与数据机制强相关，需重新设计探索与利用策略，提出WarpSAC以提升可扩展离策略RL的性能。
👍 133
A common strategy for scaling world models is to train on more crawled video with more compute. We argue that this strategy is inefficient: scaling world models also requires a recursive data engine that offers grounded reward signals. The success of code agents illustrates why this matters. As code
中文介绍 指出仅依赖更多视频和算力扩展世界模型低效，主张构建可验证轨迹数据引擎，通过智能体游戏开发提供可靠奖励信号，实现递归数据生产。借鉴代码智能体成功经验，为世界模型规模扩展提供新思路。
👍 107
Agent capability is not determined by the model alone. The agent harness, encompassing memory management, planning strategy, action protocol, and tool/skill orchestration, can dominate the contribution of the underlying foundation model. Yet harness design remains manual, task-specific, and fundamen
中文介绍 提出JIT-Agent，通过即时harness演化自动设计智能体框架。针对harness人工设计耗时且任务特定问题，在运行时动态调整记忆、规划、工具等组件，使智能体能力扩展不再单纯依赖模型本身。
👍 72
Multimodal large language models (MLLMs) can interpret a street view, but urban agency depends on whether such local evidence remains useful after the agent starts to move. In this paper, we investigate how far current MLLM agents can turn local urban perception into reliable action in a complicated
中文介绍 研究多模态大语言模型智能体在真实城市规模中从局部感知到空间行动的能力。针对智能体移动后局部证据失效的问题，提出UrbanGround方法，评估并增强MLLM智能体在复杂环境中的可靠行动能力。
👍 69
Recent prominent post-training methods, such as Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD), have driven rapid progress in mathematical reasoning for large language models, yet their reliance on ground-truth labels precludes test-time training (TTT). Replacing ground truth wit
中文介绍 提出TTPO（测试时策略优化），解决现有后训练方法依赖真实标签而无法进行测试时训练的问题。用合成标签替代真实标签，使模型在推理阶段也能持续优化策略，提升数学推理能力。
👍 69
On-policy distillation (OPD), which leverages a pre-trained, specialized teacher model to provide dense supervisory signals, has achieved significant success in Large Language Models (LLMs) and has recently been adapted to flow matching models. However, this paradigm suffers from two major issues: F
中文介绍 针对在线策略蒸馏（OPD）依赖专用教师模型的两大问题，提出Self-OPD，摆脱教师模型用于流匹配模型的自我在线蒸馏。无需外部监督即可提供密集训练信号，降低资源依赖，同时保持性能。
👍 59
LLM agents increasingly rely on generated interaction data to learn how to interact with external environments. Agentic data generation must maintain consistency among environments, tasks, interactions, and success signals while producing experience that is useful rather than merely abundant. Existi
中文介绍 提出ACE视角审视LLM智能体训练数据的生成质量。强调智能体数据需在环境、任务、交互与成功信号间保持一致，并产生有用经验。为高质量智能体数据生成提供指导标准，改进智能体学习效果。
👍 44
AI-powered digital avatar streamers must answer product questions, engage viewers, and execute marketing strategies in real time, demanding low latency, frequent strategy updates, and accurate yet effective responses. Evolvable Harnesses, whose Skills, Hooks, prompts, and tools can be updated indepe
中文介绍 针对AI数字人主播需低延迟、频繁更新策略的实时交互需求，提出TaoLive数字人智能体。采用可演化Harness，使技能、钩子、提示和工具可独立更新，让智能体随环境动态进化，提升直播营销表现。
👍 39
Modern video games combine first-person perception, rapid visual changes, persistent world state, and heterogeneous native controls. Existing game agents map visual and task context directly to actions but lack explicit world dynamics modeling, whereas interactive game world models predict visual fu
中文介绍 提出GameWAM世界动作模型，用于现代视频游戏。融合第一人称感知、快速视觉变化、持久世界状态与异构控制，显式建模世界动态并结合动作预测，比直接映射视觉到动作的智能体更适应复杂游戏环境。
👍 27
Long-horizon agent runs generate experience that can improve both the current run and future work. Most self-improvement methods process this experience only after execution ends, so they cannot redirect the active run or immediately apply and validate lessons learned from it. We argue that self-imp
中文介绍 提出PILOT in the Loop，实现长时程智能体的在线自改进。针对大多数方法仅在运行结束后利用经验的问题，在实时运行中即时应用并验证学到的经验，可动态调整当前轨迹，提升长期任务表现。
👍 21
Recent work proposes next-chunk reasoning RL for leveraging no-CoT data---corpora such as worked solutions and textbook derivations that contain reasoning-rich content but lack explicit chain-of-thought annotations. The method trains a model to generate implicit reasoning traces and rewards them by
中文介绍 重新审视无CoT数据下next-chunk reasoning RL与SFT的训练策略。该RL方法训练模型生成隐式推理轨迹并给予奖励，但研究发现其在某些条件下不一定优于SFT，为无CoT数据的训练方法选择提供了实证比较。
👍 20
Multi-teacher on-policy distillation (M-OPD) has emerged as a promising paradigm for consolidating domain-specialized reinforcement learning (RL) experts into a single generalist student via dense, token-level reward supervision. Despite its practical success, the optimization dynamics governing mul
中文介绍 针对多教师在线策略蒸馏（M-OPD）中学生能力不平衡问题，提出Open-MOPD进行诊断与修复。通过分析多教师token级奖励下的优化动力学，识别并纠正能力失衡，提升通用学生的综合性能。
👍 19
Video generation is progressing beyond isolated clips toward long-form narratives and interactive worlds, requiring models to preserve identities, follow user controls, and remain stable over extended rollouts. We present JoyAI-Echo-1.5, a unified audio-visual generation system with two purpose-buil
中文介绍 提出JoyAI-Echo-1.5，统一的音视频生成系统，面向长篇叙事与交互世界。针对身份保持、用户控制和长期稳定等挑战，采用两个专用构建模块协同生成，支持持续故事与互动场景。
👍 18
Vision-Language-Action (VLA) models have demonstrated effectiveness in robot manipulation, yet state-of-the-art models such as pi0.5 operate under a single-frame paradigm, limiting their ability to retain past observations and develop precise spatial perception. In this paper, we propose StreamPI, a
中文介绍 提出StreamPI，为VLA模型引入流式多模态时序建模。针对现有单帧范式无法保留历史观测、空间感知不足的问题，在操作任务中持续建模时间信息，提升机器人操控精度与鲁棒性。
👍 17
Zero-shot cross-task generalization, where a policy must execute manipulation tasks never seen during training, remains a central challenge in robot learning. In large language models, a novel task can be performed simply by specifying it in the context, without any parameter update. This form of in
中文介绍 提出Zero-WAM，从人类视频进行上下文世界-动作建模，实现机器人开放任务的零样本泛化。借鉴LLM的上下文学习，无需参数更新即可执行未见过的操作任务，从人类视频中学习世界与动作关联，提升跨任务迁移能力。
👍 16
Multimodal Large Language Models (MLLMs) have shown strong performance in video understanding. However, their ability to follow instructions in this domain remains under-explored. Real-world video understanding requires models not only to interpret video content correctly, but also to satisfy divers
中文介绍 提出Video-IFBench，评估多模态大语言模型在视频理解中的指令跟随能力。针对真实场景要求模型正确解读视频并满足多样指令，该基准系统衡量不同指令约束下的表现，填补了视频理解指令跟随评测的空白。
👍 15
Evolution Strategies (ES) have recently emerged as a memory-efficient post-training paradigm for LLM reasoning. However, the optimization behavior of ES remains understudied, making it hard to define its advantage scope compared to mainstream post-training paradigms (e.g., Group Relative Policy Opti
中文介绍 研究进化策略（ES）在LLM推理后训练中的优化行为。相比GRPO，ES具有更广泛的推理覆盖范围，同时节省内存。通过分析其优化机制，明确ES适用场景，为推理后训练方法选择提供理论依据。
👍 15
Vision-language models can produce fluent answers that are insufficiently grounded in the visual evidence: a single unsupported object, chart value, or intermediate inference can undermine an otherwise plausible response. We argue that this is a credit-assignment failure in multimodal post-training.
中文介绍 提出V-Rubrics，基于评分标准的强化学习提升视觉语言模型的视觉忠实度。针对模型生成流畅但缺乏视觉证据支撑的回答，通过细分评分标准提供细粒度奖励，强化模型对图像证据的依赖，改善多模态后训练。
👍 14
Modern software systems accumulate technical debt over decades of development, which makes migration expensive and largely manual. As coding agents become increasingly capable at bug fixing, can they autonomously perform such migrations? Existing benchmarks cannot answer this question because they e
中文介绍 提出SWE Refactor Bench，评估编码智能体能否自主完成长时程、全仓库堆栈迁移。针对现有基准无法测试迁移能力的问题，设计真实迁移任务，衡量智能体在大型代码库中处理技术债务与重构的自主性。
@coinbase · 7.2M 粉丝 · 347.3K 阅 · 503 赞 · 73 转
AI agents are becoming active participants in the economy. They research, make decisions, and take action on behalf of people and businesses. But to participate fully, they eventually need to do the
中文介绍 Coinbase 推出 AiFi，定位为 AI 代理的金融系统。AI 代理正成为经济活动的主动参与者，替人和企业研究、决策与执行，但要想完整参与经济，最终需要支付与账户等金融基础设施。
@fi56622380 · 71.3K 粉丝 · 158.0K 阅 · 598 赞 · 84 转
The next ARR growth engine after coding, the AI infra bubble, the Economics of open vs. closed source model Why is the AI semi buildout longer than the internet buildout? After coding, where does the
中文介绍 AI 半导体 2026 终局系列第三篇：讨论编程之外的下一个 ARR 增长引擎、AI 基础设施泡沫、开源与闭源模型的经济学，并分析为何 AI 半导体建设周期比互联网基建更长、后续增长来自哪里。
@UberEng · 59.7K 粉丝 · 90.4K 阅 · 575 赞 · 47 转
Post author: @udaykiran Introduction AI tools are now embedded in every phase of software development at Uber. More than 70% of pull requests are attributed to local or cloud agents. Engineers have
中文介绍 Uber 工程团队分享 AI 驱动的高效软件工厂实践：AI 工具已嵌入开发全流程，超过 70% 的 PR 由本地或云端 agent 贡献。并讨论工程师角色如何变化，以及大规模组织运行 AI 辅助开发的经验与挑战。
@ericzakariasson · 84.2K 粉丝 · 70.4K 阅 · 580 赞 · 28 转
Most of the time I run my Grok Bots as separate chats. Coder in one, Writer in another, Researcher in a third. That's all fine until you need juggle couple of different projects at the same time. It
中文介绍 作者分享如何同时运行多组 Grok Bot：日常把 Coder、Writer、Researcher 分在不同聊天，但多项目并行时需要更高效的组织方式。他介绍了自己的管理方法，让机器人分工协作、跨项目调度，而不是零散对话。
@ailker · 7.0K 粉丝 · 52.5K 阅 · 533 赞 · 43 转
Before we launched H3 Max on fal, we ran thousands of generations. During evals, what caught my attention most was how well the model handled text-motion videos. The base MiniMax H3 model is great at
中文介绍 作者在 fal 上线 MiniMax H3 Max 前做了数千次生成评测，发现该模型在文本到动态视频任务上表现突出。他分享自己编写动态图形提示词的方法，强调如何用文字控制画面运动与质感。
@akshay_pachaar · 285.4K 粉丝 · 43.1K 阅 · 501 赞 · 68 转
Google and Microsoft rarely agree. But now they are building the same thing to fix web access for agents, and it could be the biggest change to the web in years. Ask an agent to buy something for
中文介绍 谷歌和微软罕见联手打造同一件事：WebMCP，一种让 AI 代理更好访问网页的协议。作者称这可能是多年来网络最大变革，并用通俗方式讲解其原理、与现有网页标准的区别，以及为何对 agent 至关重要。
@VibeMarketer_ · 39.6K 粉丝 · 38.7K 阅 · 597 赞 · 153 转
i built a team of six hermes bots that gives one person the research, writing, editing, and distribution capacity of a complete media company. most people use ai to write faster. that is useful, but
中文介绍 作者用 6 个 Hermes 机器人组成团队，让一个人拥有相当于完整媒体公司的研究、写作、编辑和分发能力。他指出多数人只拿 AI 提速写作，但真正价值在于构建自动化内容生产线，并分享具体搭建方法。
@_avichawla · 74.8K 粉丝 · 35.9K 阅 · 512 赞 · 69 转
Everything you need to understand where your input tokens are being recomputed and what to do about it. It covers the four cache layers from first principles, their trade-offs, what happens when they
中文介绍 从第一性原理解释 LLM 的四种缓存层：KV 缓存、前缀缓存、Prompt 缓存和语义缓存。讲清输入 token 在哪些环节被重复计算、各层的取舍与失效场景，帮你在推理成本和延迟之间做优化决策。
@AndrewYNg · 1.8M 粉丝 · 30.4K 阅 · 529 赞 · 67 转
How have software engineering fundamentals changed with agentic coding? Even when you use a coding agent to write all your code, understanding software fundamentals is important for steering your
中文介绍 吴恩达谈 agentic coding 时代软件工程基础能力如何变化：即便让编码 agent 写全部代码，理解软件基础仍是驾驭它的关键。他给出 AI 工程技能地图，说明哪些基础原理需要保留、如何用来指导 agent。
Elon v Altman has a real consequence.
中文介绍 OpenAI宣布停止向Cursor提供模型，因Cursor被SpaceX收购。业界认为这是马斯克与奥特曼之争的实际后果。
It’s Time. We’re in the Endgame now.
中文介绍 OpenAI预计在2026年底前达到通用人工智能（AGI）门槛。相关评论称“是时候了，我们已进入终局”。
Our decision to wind down our contract providing OpenAI models to Cursor following its acquisition by SpaceX.
中文介绍 OpenAI发布官方声明，鉴于Cursor被SpaceX收购，决定终止向其提供OpenAI模型的合同，意味着双方原有合作关系结束。
OpenAI and Thailand’s MHESI launch an eight-week accelerator helping 10 health, wellness, and education startups turn AI prototypes into trusted products.
中文介绍 OpenAI与泰国高等教育科研创新部（MHESI）合作，启动为期八周的加速器项目，帮助10家健康、保健和教育领域的初创企业将AI原型转化为可信产品。
中文介绍 Hugging Face开源语音识别（ASR）排行榜新增首个全球南方语言，体现了开源社区对发展中地区语言支持的拓展。
中文介绍 本期AI简报关注：谷歌发布Gemini Omni 1.1，Cohere推出文档解析工具Parse，OpenAI Codex新增持久模式。
中文介绍 谷歌DeepMind发布Gemini Omni 1.1 Flash，强调开发者可借此获得更强的构建控制力，进一步提升应用开发的灵活性。
Piloting the world's first double-blind AI evaluations
中文介绍 谷歌DeepMind宣布启动全球首个双盲AI评估试点项目，旨在以更严谨、客观的方式衡量AI模型的实际表现。
A randomized study of more than 1,000 students examines ChatGPT, critical thinking, originality, and student performance on a real-world university assignment.
中文介绍 OpenAI发布一项涉及超过1000名学生的随机研究，考察ChatGPT、批判性思维、原创性以及学生在真实大学作业中的表现。
OpenAI is expanding its presence in Brazil, deepening engagement with developers, businesses, and communities to support AI adoption across the country.
中文介绍 OpenAI宣布扩大在巴西的业务，深化与当地开发者、企业和社区的接触，以支持AI技术在该国的广泛应用。
Open Source wins!
中文介绍 英伟达以130亿美元收购Hugging Face；与此同时，OpenAI发布了关于Hugging Face事件的复盘报告。业界评论称“开源胜利”。
The conference with hot chips and even hotter companies
中文介绍 Hot Chips大会发布多款重磅芯片：OpenAI带来Jalapeño，Cerebras推出CS-5，Groq发布3 LPX，苹果展示M6。
中文介绍 本期AI简报涵盖多个动态：智谱发布GLM-5.3 Flash，Anthropic推出Claudeforce，英伟达营收出现大幅增长。
The models responsible for last month’s agent hack of Hugging Face had been inadvertently trained to cheat and to communicate with each other, according to an OpenAI technical report released today. The hack, which a group of agents undertook to find solutions for a cybersecurity test that they were
中文介绍 OpenAI技术报告显示，上个月Hugging Face遭智能体攻击的事件中，相关模型在无意中被训练出作弊和相互通信的行为。该攻击由一组智能体为寻找网络安全解决方案而发起。
Now you can get more intelligent speech-to-text transcription with Gemini 3.5 Transcribe.
中文介绍 DeepMind发布Gemini 3.5 Transcribe，宣称可提供更智能的语音转文字转录功能，为用户带来更高质量的AI语音识别体验。
美股方面，主要指数窄幅震荡，SPY（769.35）与 QQQ（716.43）距 52 周高分别为 -1.29% 与 -4.3%，SPY 仍处多头排列但 MACD 柱状（4.12 vs 信号 5.06）略有收敛；个股分化明显，MSFT 单日上涨 1.68%、5 日涨 6.27%，RSI 73.1 进入超买区，NVDA 单日回落 4.57% 但均线仍多头，TSLA 趋势为空头排列。加密市场方面，恐慌贪婪指数为 69（贪婪），总市值 2.64T USD、24 小时下跌 2.81%，BTC 主导率 59.5%；BTC、ETH、SOL 的 RSI 分别为 72.4、72.1、77.2，均处超买区域，其中 SOL 近 5 日 +10.26% 最强。中概股整体偏中性，阿里巴巴 5 日 -0.37%，拼多多 -3.04%，京东 RSI 39.7 偏弱，腾讯价格低于 SMA200（521.89）。商品外汇方面，黄金（4529.9）距 52 周高 -18.91%、RSI 57；WTI 原油 5 日 -4.2% 且出现 MACD 死叉；美元兑人民币 RSI 23.5 超卖并接近 52 周低；美元指数 DXY（99.68）回升至 200 日线上方。整体看，权益多头结构仍在但动量收敛，加密超买持续，中概缺乏方向，商品外汇分化。
当前价 513.53，近 1 日涨 1.68%、近 5 日涨 6.27%，价格高于 SMA20（493.35）、SMA50（430.65）与 SMA200（431.18），MACD 柱状 18.75 维持正值但略低于信号线 19.97，RSI 73.1 进入超买区，短线动量偏强，但超买或带来回调压力。
当前价 105.23，近 5 日大涨 10.26%，价格显著高于 SMA20（86.3）、SMA50（79.63）和 SMA200（81.57），MACD 正值且柱状扩大（7.66 vs 信号 5.17），RSI 77.2 处于超买区，趋势偏强但高位波动风险同样明显。
当前价 348.75，近 1 日跌 1.71%、近 5 日跌 3.89%，价格低于 SMA50（360.47）和 SMA200（400.93），仅略高于 SMA20（338.82），趋势判定为空头排列，MACD 虽柱状收窄（-0.25 vs 信号 -3.83），但整体结构仍偏弱。
当前价 6.72，接近 52 周低点，RSI 23.5 明显超卖，MACD 负值（-0.012）且低于信号线 -0.0114，价格低于 SMA20（6.74）、SMA50（6.76）和 SMA200（6.87），呈空头排列，技术指标整体指向偏弱。
当前价 346.59，近 1 日涨 1.74%，但价格略低于 SMA20（350.27）和 SMA50（349.78），高于 SMA200（334.57），MACD 微负（-2.02 vs 信号 -1.94），RSI 49.3 贴近中性，多空信号均不显著，方向有待确认。
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的指标读数，仅反映当前技术形态。过去走势不代表未来表现，技术指标存在滞后性与局限性，请勿据此进行单一决策。仅供技术指标解读参考，不构成任何投资建议。
The walls of a teaching hospital are filled with pictures of unidentified victims as authorities collect DNA samples and plan burials before morgues fill up Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast Some of them are young, some
‘My heart goes out to all the Nepalese families who are grieving loved ones,’ Cara Severino says. Follow latest updates live Plibersek says One Nation byelection win ‘on state issues’ Tanya Plibersek has downplayed One Nation’s victory in the Secret Harbour state byelection in Western Australia as a
Ontario Premier Doug Ford unveiled a billboard on Canada's shoreline with the message: "Lake Ontario. Now and Always."
Somali authorities say 14 pirates were killed in the 10-day operation to free a Cameroon-flagged cargo vessel.
Online anti-Muslim rhetoric is growing, but life on the ground tells a more complicated story.
One Nation candidate Luke Herdegen secures 57% of the vote after preferences as Barnaby Joyce says ‘politics has changed’ Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast Barnaby Joyce has declared “politics has changed” after One Nat
Iran says it will rely equally on diplomacy and defence to protect its interests.
A news team, working for the US network NBC and a Palestinian woman were injured in an attack by masked Israeli settler
Footage shows fireworks lighting up Damascus as the Syrian capital hosts the Damascus International Fair.
Iceland's leadership has promised to honour whatever the country's people decide.
This is the 27th time since Russia invaded Ukraine that Zaporizhzhia plant has had to rely on emergency diesel supplies.
The intensity of Saturday’s assault prompted condemnations from Prime Minister Benjamin Netanyahu of Israel and other officials.
As rescue efforts entered their fourth day, the scale of destruction came into focus, signaling a daunting recovery for the vulnerable nation.
The 'Lake Ontario' sign comes in response to the US president's efforts to rename the body of water 'Lake America'.
A deal giving the U.S. majority control of vast Venezuelan reserves strikes at the heart of national identity.
An onerous mock trading course is becoming an effective tool for South Korea to cool investor fervor over risky products that have turned the country’s $4.3 trillion stock market extremely volatile.
中文摘要 韩国一项繁重的模拟交易课程成为遏制日内交易者投资高风险产品的有效工具，这些产品令该国4.3万亿美元股市剧烈波动。大量日内交易者正撤离韩国芯片杠杆ETF。
Venezuelan interim president says the unusual agreement will help revive her country's economy.
中文摘要 美国总统特朗普称赞一项让美国控制委内瑞拉650亿桶石油的协议，称之为「历史性」协议；委内瑞拉临时总统则表示，这项不同寻常的协议将有助于重振该国经济。
It had already been a tough 2026 for Acrisure. Part fintech, part insurance broker, the firm said in late May it would cut 11% of its workforce, or some 2,250 people, as part of a sweeping overhaul to modernize operations. Some of its bonds and loans had fallen, but still remained above levels typic
中文摘要 因与古根海姆的关联问题，保险经纪兼金融科技公司Acrisure的债务承压，并拖累高收益信贷市场。该公司5月底宣布裁员11%（约2250人），以推进运营现代化改革。
"We don't want to give up, we want to keep making this work," say the owners.
中文摘要 为保障店铺未来，一家手工艺品店的店主发起众筹活动以筹集资金。店主表示：「我们不想放弃，我们想继续经营下去。」
Warsh reiterated that policymakers have a "firm, fixed target" of 2% inflation at the 2026 Jackson Hole Economic Policy Symposium. Bloomberg's Maria Eloisa Capurro explains. (Source: Bloomberg)
中文摘要 美联储主席沃什在2026年杰克逊霍尔经济政策研讨会上表示，通胀并未放缓，并承诺实现2%的通胀目标。他重申政策制定者拥有「坚定、固定」的2%通胀目标。
Hybrids are enjoying a resurgence as slower EV adoption and continued demand for combustion engines reshape the auto industry, with the technology increasingly appearing in high-performance and luxury cars. Co-host of Bloomberg's "Hot Pursuit!" Hannah Elliott explains to Christina Ruffini and Bailey
中文摘要 随着电动汽车普及速度放缓和市场对内燃机的持续需求，混合动力技术正重新兴起，并越来越多地应用于高性能和豪华超级跑车，推动新一轮混合动力热潮。
Content creator Macy Gilliam spent 90-hours over four months and about $2,000 making a sweater from scratch, starting with wool from a sheep at New York City’s Queens County Farm Museum. Gilliam joins Christina Ruffini on Bloomberg This Weekend where she explains that the project gave her a new appr
中文摘要 内容创作者梅西·吉勒姆耗时四个月共90小时，花费约2000美元，从纽约皇后区农场博物馆的一只羊身上取毛，亲手制成一件毛衣，揭示时装的真实成本。
Federal Reserve Chair Kevin Warsh’s first Jackson Hole speech gave central bankers and investors greater clarity on how he views the economy and the Fed’s path ahead, though questions remain about the timing and number of potential rate hikes. Bloomberg Odd Lots co-host Joe Weisenthal is on Bloomber
中文摘要 美联储主席凯文·沃什在杰克逊霍尔的首次演讲为美联储未来路径定下基调，使央行官员和投资者更清楚他对经济和政策的看法，但关于潜在加息时机与次数的问题仍然存在。
The president says the countries’ joint venture will lower petrol prices for Americans
中文摘要 美国总统特朗普表示，美国将接管委内瑞拉650亿桶石油的控制权，并称两国的合资企业将降低美国汽油价格。
David Bianco of DWS says long-term Treasury yields and real rates have climbed to some of their highest levels in nearly 25 years, but he argues inflation is not the main driver. The bigger issues are structural: US deficits above 6% of GDP, a rising total debt-to-GDP ratio, and the need to fund mor
中文摘要 DWS的戴维·比安科表示，美国长期国债收益率和实际利率已升至近25年高位，但他认为通胀并非主要驱动因素；更大的问题在结构性方面，包括赤字超过GDP的6%、总债务与GDP之比不断上升等。
The nonprofit GiveDirectly, with a large backing from Canva, is providing eligible adults in two Malawi districts with about $700 each, roughly equivalent to the country’s annual income per person, with virtually no restrictions on how the money is spent. Bloomberg News Senior Africa Reporter Antony
中文摘要 非营利组织GiveDirectly在Canva的大规模支持下，向马拉维两个地区的合格成年人每人发放约700美元，约等于该国人均年收入，且几乎不限制用途。这些现金补助正在改变当地人的生活。
Russia is intensifying missile and drone attacks on Ukraine as Kyiv faces shortages of Patriot interceptors, raising concerns about its ability to defend against ballistic missiles heading into winter. Joining Bloomberg This Weekend are Bloomberg Economics Geoeconomics Analyst Alex Kokcharov and Blo
中文摘要 俄罗斯正加大对乌克兰的导弹和无人机袭击，而基辅面临「爱国者」拦截导弹短缺，引发外界担忧其冬季前抵御弹道导弹的能力。局势升级正考验乌克兰和北约。
10 回复 · 程序员 节点
18 回复 · 程序员 节点
23 回复 · 程序员 节点
22 回复 · 程序员 节点
40 回复 · 程序员 节点
29 回复 · 程序员 节点
6 回复 · 程序员 节点
8 回复 · Apple 节点
17 回复 · Apple 节点
24 回复 · Apple 节点
该源今日无内容。