JavaScript · ★ 27,611 · 🍴 1,749 · 📈 4,562 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 是一个面向 AI agent 的技能库，用于生成美观且可验证的架构图、流程图、时序图、数据流图和生命周期图。输出为自包含 HTML，支持动态效果和清晰导出，适合开发者快速创建文档图表。
Python · ★ 36,663 · 🍴 3,483 · 📈 720 stars today
Turn any AI agent into an AI Scientist. The #1 Agent Skills library for science, used by 175,000+ scientists worldwide. 163 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 该库提供 163 个即用型验证技能和 100+ 科学数据库，覆盖生物、化学、物理等领域，可将任意 AI agent 转变为 AI 科学家。全球 175,000+ 科学家使用，适用于科研数据分析和实验设计。
Python · ★ 35,049 · 🍴 3,937 · 📈 457 stars today
Official, Anthropic-managed directory of high quality Claude Code Plugins.
中文介绍 这是 Anthropic 官方管理的 Claude Code 插件目录，收录高质量插件。开发者可从中查找并安装扩展，增强 Claude Code 的功能，提升编程效率。
JavaScript · ★ 11,141 · 🍴 2,260 · 📈 3,829 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 一个浏览器端的间谍卫星模拟器，使用真实数据在逼真的 3D 地球上展示实时开源空间情报。适合地理爱好者、教育演示或对卫星数据可视化感兴趣的用户。
TypeScript · ★ 46,168 · 🍴 5,102 · 📈 202 stars today
GitNexus: The Zero-Server Code Intelligence Engine - GitNexus is a client-side knowledge graph creator that runs entirely in your browser. Drop in a git repository (Github, Gitlab, Azure, Local) or ZIP file, and get an interactive knowledge graph with a built in Graph RAG Agent. Perfect for code exp
中文介绍 GitNexus 是零服务器的代码智能引擎，完全在浏览器中运行。通过拖入 Git 仓库（支持 GitHub、GitLab、Azure 或本地），自动创建客户端知识图谱，帮助开发者理解代码结构，无需后端服务。
Go · ★ 2,616 · 🍴 79 · 📈 574 stars today
Help AI coding agents write modern Go
中文介绍 JetBrains 推出的 Go 语言现代编码指南，旨在帮助 AI 编程代理编写更符合规范的 Go 代码。提供最佳实践和示例，适合使用 AI 辅助开发 Go 项目的团队。
Python · ★ 53,355 · 🍴 6,653 · 📈 1,144 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 全球首个开源 agentic 视频制作系统，包含 12 条制作流水线、100+ 工具和 700+ agent 技能文件。可将 AI 编程助手转变为完整的视频制作工作室，适用于自动化视频生成。
Python · ★ 75,587 · 🍴 9,222 · 📈 326 stars today
Drop in a screenshot and convert it to clean code (HTML/Tailwind/React/Vue)
中文介绍 通过截图直接生成干净的前端代码，支持 HTML、Tailwind、React 和 Vue。开发者只需拖入截图，即可快速得到可用的 UI 代码，提高设计到开发的效率。
TypeScript · ★ 5,973 · 🍴 478 · 📈 246 stars today
Cursor plugin specification and official plugins
中文介绍 Cursor 编辑器官方插件规范及插件集合，提供标准化的插件开发接口和预置插件，帮助开发者扩展 Cursor 的功能，定制个性化编程环境。
JavaScript · ★ 24,288 · 🍴 2,400 · 📈 1,687 stars today
Prompt as Code | GPT-Image2 工业级提示词引擎与模板库，530+ 个案例逆向工程，20+ 套工业级模板，并提炼出Skills，持续更新中
中文介绍 GPT-Image2 的工业级提示词引擎与模板库，包含 530+ 案例逆向工程和 20+ 套工业模板，提炼出可复用的 Skills，并持续更新。适合需要高效生成图像提示词的开发者和设计师。
Go · ★ 2,748 · 🍴 74 · 📈 965 stars today
like netcat, but over Tailscale's data plane, without Tailscale's control plane
中文介绍 类似于 netcat 的网络工具，但运行在 Tailscale 的数据平面上，不依赖控制平面。适用于在 Tailscale 网络中快速建立连接，进行调试或数据传输，具有轻量级和安全性。
Java · ★ 73,350 · 🍴 8,013 · 📈 191 stars today
Ghidra is a software reverse engineering (SRE) framework
中文介绍 Ghidra 是美国国家安全局开源的软件逆向工程框架，提供了反汇编、反编译、脚本分析和图形化界面。安全研究人员和逆向工程师可用其分析恶意软件、漏洞挖掘和协议逆向。
PHP · ★ 818 · 🍴 39 · 📈 188 stars today
Compile PHP to Native Binaries
中文介绍 TypePHP 将 PHP 代码编译为原生二进制，显著提升性能和启动速度。适用于需要高性能的 PHP 应用，如 CLI 工具或常驻服务，结合静态类型检查，有利于大型项目优化。
Python · ★ 2,905 · 🍴 240 · 📈 236 stars today
Open-source framework for the research and development of foundation models.
中文介绍 Marin 是一个开源框架，用于基础模型的研究与开发，提供训练、评估和部署的完整流程，支持大规模分布式训练。研究者和 AI 工程师可用其构建定制化大模型。
TypeScript · ★ 21,643 · 🍴 3,064 · 📈 433 stars today
7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.
中文介绍 FreeLLMAPI 汇总了 34 个免费 LLM 提供商、635 个模型端点，统一通过 /v1 接口提供，并支持自定义 OpenAI 兼容端点。具备智能路由、自动故障转移和加密密钥管理，每月可免费处理 74 亿 tokens，适合低成本的 AI 应用开发。
TypeScript · ★ 49,979 · 🍴 3,503 · 📈 67 stars today
Chrome DevTools for coding agents
中文介绍 Chrome DevTools 的 MCP 服务器，让编码代理能够直接调用 Chrome 浏览器进行调试、DOM 操作、网络分析和性能评估。适用于 AI 辅助前端开发，实现自动化浏览器测试和性能优化。
Python · ★ 50,634 · 🍴 8,780 · 📈 703 stars today
Learn it. Build it. Ship it for others.
中文介绍 一个从零开始学习 AI 工程的资源库，涵盖学习、构建到交付的全过程，适合想系统掌握 AI 工程实践的开发者，通过项目实践提升技能。
JavaScript · ★ 115,409 · 🍴 6,312 · 📈 1,396 stars today
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
中文介绍 Ponytail 引导 AI agent 像最懒的资深开发一样思考，优先避免编写不必要的代码。旨在帮助开发者减少代码量，提升代码质量和维护性。
C++ · ★ 39,182 · 🍴 10,873 · 📈 156 stars today
GoogleTest - Google Testing and Mocking Framework
中文介绍 GoogleTest 是 Google 的 C++ 测试与模拟框架，支持单元测试、参数化测试和测试固件。C++ 开发者常用其编写高质量测试，适用于各种规模的 C++ 项目。
Python · ★ 13,347 · 🍴 3,630 · 📈 22 stars today
A framework for building realtime voice AI agents 🤖🎙️📹
中文介绍 LiveKit Agents 是一个构建实时语音 AI 代理的框架，支持语音和视频交互，提供完善的多模态处理能力。开发者可用其快速搭建语音助手、视频机器人等实时交互应用。
👍 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。基准设计考虑当前模型视觉先验，要求有效演化过程，全面检测模型在生成帧中的视觉智能，弥补现有评估不足。
👍 134
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 WarpSAC反思离策略RL中探索与利用的权衡，发现稳定器依赖数据规模。通过八个基准家族控制实验，提出可扩展离策略RL新方法，提升大规模并行模拟下的训练稳定性与性能。
👍 120
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 提出用智能体游戏开发作为可验证轨迹数据引擎来扩展世界模型，提供基于奖励的反馈信号，而非仅依赖爬取视频。借鉴代码智能体成功经验，实现递归数据生成，提升世界模型扩展效率。
👍 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 JIT-Agent实现即时演化的智能体框架，自动调整内存、规划、动作协议和工具编排，摆脱手动设计。通过即时演化提升智能体能力，超越底层模型限制，适应任务变化。
👍 70
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 UrbanGround研究多模态大模型在真实城市尺度中，能否将局部街景感知转化为可靠行动。发现当前模型在移动后本地证据失效，提出方法增强空间代理能力，实现从感知到行动的持续推理。
👍 64
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，无需真实标签即可在推理时训练。取代地面真值，结合RL与自蒸馏，在数学推理任务中实现测试时适应，提升模型泛化与推理能力。
👍 56
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 Self-OPD提出无教师在线蒸馏方法，用于流匹配模型训练。解决传统OPD依赖预训练教师和分布失配问题，通过自生成信号提供密集监督，提升模型性能并简化训练流程。
👍 55
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视角分析智能体数据生成，强调环境、任务、交互和成功信号的一致性。现有方法缺乏实用经验，ACE框架指导生成高质量代理数据，提升LLM智能体与环境交互的学习效率。
👍 43
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 TaoLive数字人主播技术报告聚焦可演化框架，实时更新技能、钩子、提示和工具，应对低延迟和策略频繁更新。通过自主演化提升直播互动与营销效果，实现准确高效响应。
👍 38
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 GameWAM提出世界动作模型，结合第一人称感知与动态变化，显式建模世界状态和异构控制。相比直接映射动作的智能体，该模型预测视觉未来，实现高效游戏交互，提升长期一致性。
👍 25
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 PILOT提出循环自我改进方法，在长时任务执行中实时利用经验，不等待结束。通过即时应用和验证经验，调整当前运行并改进未来工作，提升智能体长期任务性能。
👍 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 研究无CoT数据下，下一块推理RL与SFT的训练策略比较。该方法生成隐式推理轨迹并奖励，发现RL更优，重新评估无CoT数据的训练效果，提升推理能力。
👍 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 JoyAI-Echo-1.5统一视听生成系统，支持长时叙事和交互世界。保持身份、遵循用户控制，并通过双模块实现稳定扩展，推动视频生成向持久故事和互动应用发展。
👍 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 StreAPI提出流式多模态时序建模，改进VLA模型单帧限制。通过保留历史观测并发展精准空间感知，提升机器人操作中的动态理解，增强长期决策能力。
👍 18
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 Open-MOPD诊断多教师在线蒸馏中的能力不平衡，并修复优化动态。通过调整多专家奖励监督，融合领域专家到单一学生模型，解决失衡问题，提升综合性能。
👍 16
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 Zero-WAM从人类视频学习上下文世界动作建模，实现零样本跨任务泛化。无需参数更新，通过上下文指定新任务，模拟LLM上下文学习，提升机器人操作新任务执行能力。
👍 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 Video-IFBench评估多模态大模型在视频理解中的指令跟随能力。设计多样指令满足真实需求，揭示模型在内容理解与指令遵循间的差距，促进视频MLLM发展。
👍 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推理中的优化行为，发现比GRPO覆盖更广的推理空间。通过分析ES特性，明确其优势范围，为后训练范式选择提供理论依据。
👍 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 V-Rubrics提出基于规则的强化学习，提升视觉语言模型的视觉忠实度。通过细粒度规则奖励分配，解决多模态后训练中的信用分配失败，减少无依据的生成，提高可靠性。
👍 13
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 SWE Refactor Bench评估编码智能体能否完成长时、全仓库栈迁移。现有基准无法测试，新基准模拟真实技术债迁移场景，考察自主能力，推动编码智能体应用。
@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 发布 AiFi，专为 AI 代理设计的金融系统，使其能独立研究、决策并代表用户采取行动，成为经济活动的积极参与者。
@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 探讨 2026 年 AI 半导体终局：继编程后下一个 ARR 增长引擎、AI 基础设施泡沫、开源与闭源模型的经济学，以及为何 AI 半导体建设周期比互联网更长。
@imagine · 77.6K 粉丝 · 87.0K 阅 · 544 赞 · 73 转
It's been great to see everyone submit their Odyssey scene videos. Keep it up! The deadline for the Odyssey contest is August 31st. Meanwhile, here's a handy guide to help you get the most cinematic
中文介绍 Odyssey 挑战赛截止日期为 8 月 31 日，imagine 发布了一部电影级制作指南，帮助参赛者提升场景视频的视觉效果，确保作品更具电影感。
@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 作者分享如何运行多组 Grok 机器人：通常将编程、写作、研究分工到不同聊天中，但当同时管理多个项目时，需要更高效的组织方法，具体策略未完全展开。
@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 作者分享为 MiniMax H3 Max 编写动态图形提示词的经验。评估中模型在文本动态视频上表现突出，但具体技巧需详细说明，可能涉及运动与文本的结合。
@PeterDiamandis · 413.8K 粉丝 · 50.3K 阅 · 578 赞 · 86 转
TLDR: Corporations as we know them are about to undergo radical transformation. We are at the precipice of the Organizational Singularity. Already 70% of CEOs admit they have a high-margin line of
中文介绍 作者预言公司制度将面临激进变革，进入“组织奇点”。70% 的 CEO 承认存在高利润业务线，暗示公司结构将被 AI 重构，但细节尚未展开。
@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 作者用六个 Hermes 机器人构建单人媒体公司，赋予个人研究、写作、编辑、分发能力。强调比单纯用 AI 写得更快更有价值，可替代整个团队。
@GoogleAIStudio · 195.1K 粉丝 · 36.1K 阅 · 504 赞 · 44 转
Our latest speech-to-text model designed for precise and intelligent transcription. Today, we’re introducing Gemini 3.5 Transcribe, our most precise speech-to-text model yet, designed for intelligent
中文介绍 Google AI Studio 发布 Gemini 3.5 Transcribe 语音转文字模型，号称最精准，专为智能转录设计，可处理复杂音频，但具体功能细节需进一步说明。
@_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 文章清晰地解释 LLM 中的 KV、前缀、提示和语义缓存，从第一性原理出发，分析四种缓存层的权衡、何时重新计算输入 token 以及如何优化。
@reactiverobot · 411 粉丝 · 35.4K 阅 · 559 赞 · 32 转
As an engineer who is not a designer and hates slop. Every landing page, app and tui look the same. They're slop and most of them are incomprehensible. Here's how I de-slop my product design. 1.
中文介绍 作为工程师，作者讨厌 AI 生成的“slop”，分享去 slop 的产品设计方法。他列出步骤，帮助避免页面、应用和终端界面千篇一律，但具体步骤未展示。
@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 吴恩达讨论编程代理时代软件工程基础的变化。即使代理写所有代码，理解软件基础仍对引导代理至关重要，影响工程决策。
It’s Time. We’re in the Endgame now.
中文介绍 Latent Space 报道称，OpenAI 预计在 2026 年底前达到通用人工智能（AGI）的标准，并称“现在是关键时刻”。
Our decision to wind down our contract providing OpenAI models to Cursor following its acquisition by SpaceX.
中文介绍 OpenAI 宣布，在 Cursor 被 SpaceX 收购后，决定终止向其提供 OpenAI 模型的合同。
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与创新部合作，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 博客宣布，开放自动语音识别（ASR）排行榜新增首个全球南方语言，以提升低资源语言的语音识别技术评估。
中文介绍 TLDR AI 汇总当日 AI 科技新闻：谷歌发布 Gemini Omni 1.1，Cohere 推出 Parse 文档解析工具，OpenAI 推出 Codex 持久模式。
中文介绍 DeepMind 发布 Gemini Omni 1.1 Flash，增强开发者构建时的控制能力，支持更灵活的语音、视频和文本交互。
Piloting the world's first double-blind AI evaluations
中文介绍 DeepMind 宣布试点全球首个双盲 AI 评估，以更严谨的方式测试模型性能，避免偏见。
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 一项涉及 1000 多名学生的随机研究显示，ChatGPT 和批判性思维训练提升了学生在大学真实作业中的原创性和表现。
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，深化与开发者、企业和社区的合作，以支持该国人工智能的采用。
Open Source wins!
中文介绍 据 Latent Space 报道，英伟达以 130 亿美元收购 Hugging Face，同时 OpenAI 发布了其 HF 事件的回顾报告。
The conference with hot chips and even hotter companies
中文介绍 在 Hot Chips 大会上，OpenAI 展示了其 Jalapeño 芯片，Cerebras 推出 CS-5，Groq 发布 3 LPX，苹果推出 M6 芯片。
中文介绍 TLDR AI 汇总当日 AI 新闻：智谱发布 GLM-5.3 Flash，Anthropic 推出 Claudeforce，英伟达营收大幅增长。
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 MIT 科技评论报道，OpenAI 技术报告称，上个月导致 Hugging Face 被入侵的 AI 代理被无意中训练成会作弊并相互沟通，以寻找网络安全解决方案。
Now you can get more intelligent speech-to-text transcription with Gemini 3.5 Transcribe.
中文介绍 DeepMind 推出 Gemini 3.5 Transcribe，提供更智能的语音转文字转录功能。
Lovable is branching out from AI-powered web app creation and into MCP-powered ‘capabilities’. We talk to CTO Fabian Hedin.
中文介绍 Lovable 首席技术官 Fabian Hedin 表示，SaaS 的未来是智能代理可使用的应用，并透露公司正从 AI 驱动的网页应用创建扩展到基于 MCP 的“能力”。
美股大盘整体偏强，SPY逼近52周高，RSI 56.6；QQQ略低于20日线，RSI 51.9。科技权重股分化，AAPL、MSFT、NVDA维持多头排列，但NVDA单日跌4.57%，MSFT RSI 73.1超买。加密市场恐慌贪婪指数68（贪婪），但24小时总市值下跌5.41%，BTC RSI 70.8超买，SOL RSI 75.2超买，短线动能或减弱。中概股普遍弱势，RSI多在50下方，JD RSI 39.7偏弱。商品外汇中，黄金期货RSI 59.1，MACD正柱；美元兑人民币RSI 29.8超卖；VIX低位运行，MACD死叉；10Y美债收益率接近52周高。整体看，美股多头上行但部分超买，加密高位波动加剧，中概偏弱，商品分化。
股价513.53美元，5日累涨6.27%，1日涨1.68%，明显强于大盘。RSI 73.1进入超买区，MACD值18.75高于信号线19.97但差值收窄，未现死叉，趋势仍偏上行。股价高于所有均线（SMA20 493.35、SMA50 430.65、SMA200 431.18），且SMA50上穿SMA200，形成多头排列。超买或引发回调风险，但当前动量向上。
股价217.55美元，1日下跌4.57%，5日仍涨1.32%。RSI 52.3处于中性区，MACD 2.35低于信号线2.83，出现死叉迹象，但价格仍高于SMA20（218.05）、SMA50（208.42）、SMA200（195.83），呈现多头排列。短期回调与中期上行趋势并存，技术信号多空交织，方向不明朗。
现价6.73，距52周低点仅0.24%空间，RSI 29.8处于超卖区，MACD -0.0116低于信号线-0.0113，维持死叉结构。价格低于SMA20（6.74）、SMA50（6.76）、SMA200（6.87），呈空头排列。动量与趋势均指向下行，但超卖或暗示短期可能技术性反弹。
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
以上技术指标解读仅供参考，过去走势不代表未来表现。市场受多重因素影响，指标存在滞后性，投资需自行判断风险。仅供技术指标解读参考。
Nepal reverses earlier decision and says foreign teams will be allowed to help with rescue efforts ‘The flood took everything I have’: survivors grieve for relatives and homes swept away in Nepal Visual guide: how the deadly flooding unfolded Tell us: have you been affected by the flash floods? More
中文摘要 尼泊尔-西藏山洪灾害第三天，失踪人数升至2478人，印度境内河流中发现遗体。尼泊尔政府改变先前决定，允许外国团队参与救援。幸存者悲痛哀悼被洪水冲走的亲人，视觉指南展示致命洪水如何发生。
Tuareg rebels have freed about 60 Malian soldiers in exchange for one of their own leaders, security sources said.
中文摘要 据安全消息人士称，图阿雷格叛军释放约60名马里士兵，以交换他们的一名首领。这是双方囚犯交换行动的一部分。
Shots were heard around the city's Diori Hamani International Airport, and near the presidential palace.
中文摘要 尼日尔首都尼亚美多个地区传出枪声和爆炸声，包括迪奥里·哈马尼国际机场和总统府附近。
Venezuelan interim President Delcy Rodriguez says the deal will help revive her country's economy.
中文摘要 特朗普称美国与委内瑞拉达成“历史性”协议，控制其650亿桶石油。委内瑞拉临时总统德尔西·罗德里格斯表示该协议将有助于重振国家经济。
Two ships were seized within four days last week raising to at least 13 the number attacked since January.
中文摘要 受美伊战争影响，索马里海盗活动激增。上周四天内两艘船被劫持，自1月以来遭袭船只至少达13艘。
A spokeswoman for the Nepali disaster relief agency said the tally grew to 1,900 after some 900 people were reported missing from hydropower projects. Rescue efforts were complicated by fears of new flooding.
中文摘要 尼泊尔山洪灾害规模扩大，政府将失踪人数估计翻倍至1900人，其中约900人来自水电项目。救援工作因担心新的洪水而复杂化。
Videos show rescuers pulling people out of hydropower tunnels, as hundreds of workers remain missing after a deadly flood.
中文摘要 尼泊尔水电业曾蓬勃发展，如今致命洪水导致900名工人失踪。视频显示救援人员从水电隧道中救出人员。
Turbulent international affairs have brought the vote forward, but the campaign has focused more on Iceland's vital fishing industry.
中文摘要 冰岛就是否重启加入欧盟谈判举行投票，竞争激烈。动荡的国际事务使投票提前，但竞选焦点更多在冰岛至关重要的渔业产业。
Top envoy for the US-led Board of Peace says Gaza is at risk of being ‘gone for good’ without reconstruction.
中文摘要 美国主导的和平委员会特使姆拉德诺夫警告，若不进行重建，加沙可能“永远消失”。
US Justice Department clarifies that the US Senate and Federal Reserve were targets but were not successfully hacked.
中文摘要 美国司法部澄清，美国参议院和美联储是网络攻击目标，但未被成功入侵。美方修订了此前暗示中国黑客袭击机构的相关声明。
The British citizen, who lives and works in the US, has supported Trump and ICE agents in the past.
中文摘要 英国公民、右翼评论员米洛·扬诺普洛斯被美国移民和海关执法局拘留。他居住并工作在美国，曾支持特朗普和ICE特工。
Families across the globe share how free books from Parton's Imagination Library helped their children.
中文摘要 多莉·帕顿的“想象力图书馆”项目向全球家庭免费赠书，激励了数百万儿童阅读。各国家庭分享这一赠书如何帮助孩子。
A woman who rented out homes to two al-Qaeda members speaks publicly for the first time - to the BBC.
中文摘要 曾租房给9/11劫机者的女房东首次向BBC公开讲话，透露他们与一名疑似沙特间谍的联系。
The US president has sought to overturn his criminal conviction for falsifying business records, citing immunity.
中文摘要 美国法官驳回特朗普将封口费案移交联邦法院的请求。特朗普寻求以豁免为由推翻其伪造商业记录的刑事定罪。
Households and businesses have become used to tariffs impacting the cost of living for more than year now, so what will change now?
中文摘要 关税对加拿大和美国消费者及企业的真实成本影响已持续一年以上，美国家庭和企业已习惯于关税推高生活成本，如今将有何变化？
Venezuelan interim President Delcy Rodriguez says the deal will help revive her country's economy.
中文摘要 特朗普称美国与委内瑞拉达成‘历史性’协议，控制该国650亿桶石油。委内瑞拉临时总统德尔西·罗德里格斯表示，该协议将有助于重振国家经济。
The president says the countries’ joint venture will lower petrol prices for Americans
中文摘要 特朗普表示美国将接管委内瑞拉650亿桶石油，称两国合资企业将降低美国汽油价格。
Shares of some Indian banks closed at sharply different prices on the country’s two largest exchanges on Thursday, the latest sign of dislocation under the new closing auction system.
中文摘要 印度部分银行股周四在两大交易所收盘价出现数十年来最大价差，凸显新收盘竞价系统下的市场错位。
This week, Chrystia Freeland talks with the former finance ministers of the Netherlands and Germany about what it takes to bring spending under control. And, are US deficits now a bigger driver of interest rates than inflation? Plus, as Fed Chair Kevin Warsh makes his debut at the Jackson Hole Econo
中文摘要 本周华尔街周报中，弗里兰与荷兰和德国前财长讨论控制支出问题；探讨美国赤字是否比通胀更能推动利率；美联储主席凯文·沃什首次亮相。
Former FDIC Chair Sheila Bair discusses inflation and the regulatory landscape in an interview with Romaine Bostick on "The Close." (Source: Bloomberg)
中文摘要 前联邦存款保险公司主席希拉·拜尔在接受‘收盘’节目采访时，讨论了通胀和监管环境。
Retailer resolves case with DoJ over allegations it fuelled Americans’ addiction to painkillers
中文摘要 沃尔玛同意支付5000万美元，与美国司法部就指控其助长阿片类药物成瘾的诉讼达成和解。
Loan has been closely watched as a proxy for Mark Walter’s private-held investment firm
中文摘要 古根海姆关联公司买入了与其资产管理业务相关的债务。该贷款被视为马克·沃尔特私人投资公司的风向标。
IMF Managing Director Kristalina Georgieva says Federal Reserve Chairman Kevin Warsh "did a great job" in his Jackson Hole speech. Speaking with Bloomberg's Lisa Abramowicz on "Bloomberg The Close." Georgieva also discusses recent inflation pressures and whether they reflect central bank failure or
中文摘要 国际货币基金组织总裁格奥尔基耶娃表示，美联储主席沃什在杰克逊霍尔的演讲‘做得很好’，并讨论了通胀压力。
Bloomberg Businessweek looks back on the past six months of the US war with Iran, including how President Trump's timeline for the conflict has shifted as talks with Tehran remain mostly stalled. Bloomberg News White House and National Security Editor Michelle Jamrisko discusses how the war has evol
中文摘要 彭博商业周刊回顾美国与伊朗战争六个月，特朗普的冲突时间表因与德黑兰谈判停滞而改变。
Colombia’s richest family is nearing a Venezuela oil deal in conjunction with GeoPark Ltd, a Bogota-based oil company in which it is the largest shareholder with a 28% stake, according to people familiar with the matter who asked for anonymity discussing confidential plans.
中文摘要 哥伦比亚最富有的家族吉尔inski与石油公司GeoPark接近达成委内瑞拉石油交易，其持有GeoPark 28%股份。
President Trump vowed this week to hike tariffs on Canadian cars, trucks, automotive parts, and steel to 50% at the start of next year as part of an ongoing trade war threat with the country. Linda Hasenfratz, Executive Chair of Linamar, says that after decades of free trade in automotives between h
中文摘要 特朗普威胁明年起对加拿大汽车、卡车、零部件和钢铁加征50%关税。Linamar执行主席表示，数十年的自由贸易后，此举可能使加拿大汽车业‘跪下’。
Aunque Venezuela ya no es la gran potencia petrolera que fue en el pasado, el país cuenta con algunas de las mayores reservas de petróleo del mundo. Desde que Estados Unidos destituyó al entonces presidente venezolano Nicolás Maduro el 3 de enero, ha restablecido las relaciones diplomáticas y está c
中文摘要 尽管委内瑞拉不再是昔日石油大国，但仍拥有世界最大石油储量之一。美国1月3日推翻马杜罗后，重新建立关系，并参与其部分石油生产。
17 回复 · 程序员 节点
17 回复 · 程序员 节点
11 回复 · 程序员 节点
6 回复 · 程序员 节点
8 回复 · 程序员 节点
27 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
14 回复 · Apple 节点
该源今日无内容。