JavaScript · ★ 18,196 · 🍴 1,871 · 📈 1,698 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 面向GPT-Image2的工业级提示词引擎与模板库，通过逆向工程530+案例提炼可复用Skills，内置20+套生产级模板，帮助开发者以Prompt as Code的方式高效生成图像。适合需要批量、稳定调用图像生成能力的工程师与设计师。
Python · ★ 1,791 · 🍴 178 · 📈 351 stars today
Community plugin marketplace for Claude Cowork and Claude Code. Read-only mirror — submit plugins at clau.de/plugin-directory-submission.
中文介绍 Claude Cowork与Claude Code的社区插件市场，收录并展示社区贡献的插件，采用只读镜像方式托管，插件提交入口指向官方目录。帮助用户扩展Claude工具链能力，开发者可通过该仓库发现、评估和使用各类插件。
Python · ★ 35,368 · 🍴 12,157 · 📈 1,265 stars today
The job search that runs on your machine. AI job application framework built on Claude Code: evaluate postings, tailor CVs, write cover letters, prep interviews. Fork it and own it.
中文介绍 基于Claude Code的本地AI求职框架，运行在自己机器上，自动评估职位信息、定制简历、撰写求职信并准备面试。整个流程可自行fork掌控，适合注重隐私、想个性化定制求职流程的开发者。
TypeScript · ★ 3,369 · 🍴 332 · 📈 543 stars today
Apache Maka (Incubating) is a local-first AI agent workspace. Model messages, tool calls, tool results, permission decisions, and termination events are recorded as an append-only log.
中文介绍 Apache Maka（孵化中）是一款本地优先的AI Agent工作区，将模型消息、工具调用、工具结果、权限决策与终止事件全部记录为追加式日志，便于审计和回溯。适合需要可追溯、可复现agent行为的开发者和研究团队。
Python · ★ 100,298 · 🍴 19,355 · 📈 218 stars today
TradingAgents: Multi-Agents LLM Financial Trading Framework
中文介绍 基于大语言模型的多智能体金融交易框架，模拟多个专业角色协作进行市场分析、讨论并做出交易决策。适合量化研究和金融AI实验，提供完整的agent协作流程与决策框架。
Python · ★ 12,769 · 🍴 1,381 · 📈 813 stars today
Self-organizing AI second brain for Obsidian + Claude Code. Drop any source and Claude reads, links, and files it into one connected knowledge graph of plain Markdown you own. AI note-taking, personal knowledge management (PKM), and an open-source Notion alternative. Based on Karpathy's LLM Wiki pat
中文介绍 将Obsidian与Claude Code结合，打造自组织的AI第二大脑。拖入任意资料，Claude自动读取、关联并归档为纯Markdown知识图谱，实现个人笔记与资料的自动化管理。适合Obsidian用户和知识管理爱好者。
Python · ★ 49,047 · 🍴 8,571 · 📈 569 stars today
Learn it. Build it. Ship it for others.
中文介绍 一份从零开始的AI工程学习路线，强调「学、造、交付」三步走。涵盖理论到实践，引导学习者亲手构建并向外发布AI应用。适合希望系统入门AI工程、动手实现项目的开发者。
Rust · ★ 37,800 · 🍴 3,737 · 📈 542 stars today
Your Personal AI super intelligence. A brain that builds a local-first memory of your life, a fantastic orchestrator of agent fleets and workflows, and a deep researcher.
中文介绍 构建本地优先的个人记忆库，并作为编排器调度多个agent与工作流，还能进行深度研究，旨在成为你的个人AI超级智能。适合追求数据自主、需要个性化智能助手的用户。
Shell · ★ 31,327 · 🍴 3,182 · 📈 1,083 stars today
Beautiful, Modern & Opinionated Linux
中文介绍 一款注重美观与现代感的Linux发行版，采用「有主见」的默认配置，开箱即用。适合追求界面精致、希望减少配置时间的开发者和桌面用户。
Python · ★ 134,257 · 🍴 19,744 · 📈 161 stars today
100+ AI Agents, Agent Skills and RAG Apps - Free and Open Source.
中文介绍 收录100多个免费开源的AI Agent、Agent Skills与RAG应用，覆盖多种场景。为开发者提供可直接参考、复用和部署的LLM应用示例，是快速上手agent开发与检索增强生成的资源库。
★ 207,252 · 🍴 21,146 · 📈 830 stars today
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
中文介绍 基于Andrej Karpathy关于LLM编码陷阱的观察，提炼为单个CLAUDE.md配置文件，用于优化Claude Code的默认行为。帮助开发者规避常见问题，提升AI辅助编程的效率与质量。
Rust · ★ 118,171 · 🍴 18,004 · 📈 1,181 stars today
Lightweight coding agent that runs in your terminal
中文介绍 OpenAI开源的轻量级编码代理，直接在终端中运行，可理解代码库并自动完成编程任务。适合开发者快速执行代码修改、测试与维护，在命令行工作流中无缝集成AI辅助。
Python · ★ 2,136 · 🍴 195 · 📈 231 stars today
Open-source framework for the research and development of foundation models.
中文介绍 面向基础模型研究与开发的开源框架，提供从数据、训练到评估的完整工具链，支持研究人员和工程师构建、实验和迭代大规模模型。适合AI实验室与高校研究团队。
JavaScript · ★ 111,130 · 🍴 6,106 · 📈 982 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 通过提示词工程引导AI Agent模仿资深开发者的「懒惰」思维——优先采用最简单、不写不必要代码的方案。帮助减少过度工程，生成更有针对性、更简洁的代码。适合追求代码精简的团队。
Python · ★ 34,108 · 🍴 3,869 · 📈 55 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，收录经过审核的插件，方便用户发现和安装官方认可的扩展。确保插件安全可靠，增强Claude Code的功能与生态。
Go · ★ 2,812 · 🍴 127 · 📈 98 stars today
Your own search engine
中文介绍 可自托管的个人搜索引擎，让你完全掌控搜索索引和结果。通过自己部署，获得无跟踪、隐私友好的搜索体验，并可按需定制。适合注重隐私、希望摆脱大型平台搜索依赖的用户。
👍 173
General-purpose language models can reason and synthesize knowledge, but complex work also requires sustained interaction with files, information sources, and executable code, together with state maintenance, failure recovery, and verifiable delivery. We call this working capability: sustained, veri
中文介绍 提出 Apodex 1.1，聚焦大模型在复杂工作中的“工作能力”——对文件、信息源和可执行代码的持续交互，以及状态维护、失败恢复和可验证交付，使模型不仅能推理，还能完成多步骤实际任务。
👍 55
E-commerce live streaming requires omni-modal understanding of noisy, temporally extended streams, where product facts are distributed across speech, video frames, product images, overlaid text, and user queries. We present TLive-Omni, an omni-modal understanding model tailored to live-commerce scen
中文介绍 提出 TLive-Omni，面向电商直播的全模态理解模型，融合语音、视频帧、商品图像、叠加文本和用户查询，在嘈杂且时序延长的直播流中抽取商品事实，服务实时导购与问答场景。
👍 34
As on-device LLM agents evolve into personal copilots, the mobile operating system has become a key testbed for this paradigm, making rigorous capability evaluation essential. Yet existing benchmarks fall into two camps, each with a critical blind spot: GUI-centric benchmarks test surface-level scre
中文介绍 提出 MobilePA-Bench，面向移动端规划智能体在复杂真实任务上的能力评测。现有基准分 GUI 导向与任务导向两类，各有盲区；该基准强调对系统状态、应用内操作和长期目标的综合规划。
👍 33
Parallel reasoning improves the accuracy and robustness of large reasoning models by exploring multiple solution paths, but its computational cost grows with reasoning depth and branch count. Existing methods for managing these parallel paths typically rely on final-answer consensus, local token con
中文介绍 提出 ParaTempo，用时间置信度调度并行推理路径，替代依赖最终答案共识或局部 token 置信度的路径管理方法，在保留大推理模型多路径探索收益的同时，降低随推理深度和分支数增长的计算成本。
👍 32
Language models are sequential processors, but long-horizon agency requires external information and computation beyond model weights and active context. Prime Agent is an open-source harness for long-horizon evaluation and coding-agent workflows. A persistent IPython REPL follows the Recursive Lang
中文介绍 Prime Agent 是一个开源的长周期智能体评测与编码工作流 harness，通过持久化 IPython REPL 并遵循递归语言模型机制扩展上下文，支持自我改进，适合需要持续交互、外部计算和可靠交付的智能体任务。
👍 28
Recent omni-modal large language models (Omni-LLMs) show great potential as real-time video assistants, which continuously perceive environments and guide users to achieve specific goals. Unlike traditional passive video understanding, interactive assistants should actively combine visual states, us
中文介绍 提出 OmniAssistBench，面向 omni-LLM 的助手式交互基准。与被动视频理解不同，它评测模型在实时视频场景中持续感知环境，结合视觉状态和用户指令主动引导用户完成目标的能力。
👍 27
While text-to-3D generation has advanced rapidly, achieving high geometric fidelity at low inference cost remains challenging. Existing text-to-3D methods either decode discrete shape tokens autoregressively or iteratively refine global 3D representations with diffusion or flow models. However, auto
中文介绍 提出 Block3D，通过分块扩散实现文本到 3D 生成，避免自回归离散 token 或全局迭代扩散的高推理成本，在保持几何保真度的同时提升生成效率。
👍 15
Policy optimization (PO) for Large Language Models faces a stability--exploration trade-off, currently mediated by an action-side Policy-KL regularizer. This puts practitioners in a double bind: keeping Policy-KL constrains response behavior and consumes the action-side exploration budget, while dro
中文介绍 针对 LLM 策略优化中稳定性和探索的两难，提出环境正则化方法：不再单靠 action-side Policy-KL 约束响应行为，而是对环境侧进行正则，释放被占用的探索预算，同时维持策略稳定性。
👍 15
Open-ended real-world interaction admits multiple valid behaviors: an agent may answer directly, ask for clarification, provide progress updates, or confirm before acting. This flexibility breaks a core assumption behind group-based RL: rollouts compared within a group are no longer guaranteed to be
中文介绍 提出 ARC，面向开放式真实交互中多种合法行为并存的情况，解决 group-based RL 因组内 rollout 不可直接比较而失效的问题，实现更公平的相对优势比较，提升策略优化信号质量。
👍 15
On-policy distillation (OPD) transfers teacher capabilities by supervising trajectories sampled from the student's own policy, yet its generalization behavior remains poorly understood, as most studies evaluate OPD on a single domain and on benchmarks close to the training data. We present a control
中文介绍 通过对照实验揭示 on-policy distillation 的泛化双面性：模型可能获得向训练分布外迁移的正向泛化，也可能因学生轨迹分布偏差产生负面泛化；强调 OPD 评估需覆盖多域和分布外基准。
👍 13
Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external
中文介绍 提出 WorldMind，将游戏世界模型与 NPC 行为解耦，使 NPC 行为由显式世界状态驱动，而非隐式纠缠在视频生成或外部脚本中，从而实现状态感知、动作可控的游戏角色行为。
👍 12
When a long-horizon agent execution fails, outcome-level evaluation reveals the unsuccessful result but not where the decisive error entered the trajectory. Developers must then inspect the full execution to identify the responsible role and localize the earliest decisive root-cause step. Existing f
中文介绍 提出 LongRCA Bench，用于诊断长周期智能体执行失败：在结果级评估之外，自动定位最早的决定性错误步骤，并判定责任角色，减少开发者逐段检查完整轨迹的成本。
👍 12
Recent large language models (LLMs) can operate as coding agents that build complete games from natural language requests. Game development is especially demanding because program logic, visual and audio content, interfaces, interaction and playability must function together in one executable artifa
中文介绍 提出 GameXpert-Bench，评测 coding agents 从自然语言需求构建完整游戏的程度，覆盖程序逻辑、视觉音频内容、界面、交互与可玩性的整体集成，衡量其接近专家级游戏开发的水平。
👍 9
Serving large language models cheaply increasingly means shipping models that are both structurally compressed to a fraction of their parameters and quantized to 4 bits. Together these steps degrade reasoning, mathematics, coding, and long-context behavior enough to require a recovery, or healing, s
中文介绍 提出量化感知修复方法，恢复经结构压缩和 4-bit 量化后的 LLM 在推理、数学、代码和长上下文任务上的性能下降，为低成本模型部署提供实用修复流程。
👍 9
Recent agent benchmarks increasingly ground evaluation in executable environments, from code repair to web navigation, app APIs, and function calling. Yet completing consequential work beyond code requires more than producing a plausible response or valid tool call: agents must gather missing inform
中文介绍 提出 Thinkingbox，一个有状态业务工作流智能体的沙箱与评测基准。它强调可靠完成多步骤任务不仅需要生成文本或工具调用，还要主动收集缺失信息、维护状态，并以多次成功而非单次成功衡量可靠性。
👍 9
Agents learn to act through interaction with environments, yet the environments used for training are often manually constructed or synthesized around predefined tasks and benchmarks. This task-centric paradigm makes it difficult to scale environments that reflect realistic and evolving workflows wh
中文介绍 提出 AgentMercury，让智能体大规模合成可验证的业务场景环境，突破人工构建或围绕预定义任务生成的 task-centric 局限，为训练和评测适应现实演化工作流的智能体提供可扩展环境。
👍 8
Autonomous research systems are increasingly capable of executing long research workflows, yet automation alone does not ensure that the resulting process remains scientifically grounded. We introduce AutoResearch, a two-stage system that connects Idea Generation with Idea Execution to address both
中文介绍 提出 AutoResearch 两阶段系统，将想法生成与想法执行相连，确保自主研究系统在长流程自动化执行中保持科学 grounding，减少幻觉，实现从洞察到可验证研究产出的闭环。
👍 8
On-policy distillation (OPD) has emerged as an effective framework for post-training language models by pairing student-generated trajectories with dense token-level supervision from a teacher. However, OPD implicitly assumes that teacher-derived rewards are an appropriate proxy for reasoning progre
中文介绍 提出按推理进展过滤 on-policy distillation 训练样本的方法，不再把教师奖励直接视为推理进步信号，而是基于学生轨迹中的实际推理进展筛选监督，减少对教师行为的简单模仿。
👍 7
We present a novel approach to efficient LLM harness optimization through adaptive validation task selection. Harness optimization iteratively rewrites the harness code based on validation performance, enabling substantial performance gains without updating the underlying model weights. Existing app
中文介绍 提出 Task-CoEvolve，通过自适应验证任务选择优化 LLM harness 代码。它迭代改写 harness 而不更新模型权重，并智能挑选验证任务，以更少评估成本取得更大的性能提升。
👍 7
Small language models are usually built like large ones and then squeezed onto a CPU afterwards. We did the opposite: we fixed the target first, one user, one token at a time, 4-bit weights, ordinary CPU, and chose the architecture to suit it. The result keeps full attention in only 6 of its 18 bloc
中文介绍 Daedalus-150M 是为 CPU 推理设计的 150M 参数卷积-注意力混合模型：以单用户逐 token、4-bit 权重和普通 CPU 为约束，18 个 block 中仅 6 个保留完整注意力，兼顾效率与质量。
@chddaniel · 25.5K 粉丝 · 299.3K 阅 · 532 赞 · 26 转
69.8M impressions. 321K likes. 319K bookmarks. My brother David and I run these 3 X accounts as one content system. My Grok Bot is behind the ideas, the posts, the replies - everything. One post got
中文介绍 用 Grok Bot 管理 3 个 X 账号，组成统一内容系统，实现 6980 万曝光、32.1 万赞、31.9 万收藏。Grok Bot 负责选题、发帖、回复全部环节，团队是博主与兄弟两人。展示 AI 驱动账号矩阵的实战效果。
@P_Bonnet · 5.7K 粉丝 · 262.7K 阅 · 519 赞 · 74 转
There are countless takes telling us why the AI buildout will result in a bubble that will collapse under its own weight. Some argue we are already in it, and that the subsequent meltdown will be
中文介绍 针对 AI 建设必成泡沫的主流论调，作者以铝业贸易历史为镜，探讨「如果一切顺利」：AI 基础设施投资可能带来长期繁荣而非崩塌，正如铝从贵金属变为大宗商品。提供反共识的长期视角。
@johnbai · 11.7K 粉丝 · 181.3K 阅 · 565 赞 · 29 转
At SpaceXAI, we’re building agents that can take an idea and begin making it real. For designers, that changes what’s worth exploring. Design has always been a process of giving form to an idea:
中文介绍 SpaceXAI 团队正在构建能让想法落地的 AI 代理。作者从设计师角度，探讨 AI 如何改变「什么值得探索」，并用 Grok Bot 设计 Grok Bot，展现 AI 时代设计流程的自我迭代与可能性。
@harsehaj · 3.3K 粉丝 · 135.1K 阅 · 507 赞 · 35 转
i proposed and executed an engineering project end-to-end this summer as a growth engineering intern at @browserbase: blogEO, an engine that audits, rewrites, and generates SEO/AEO-tailored blogs on
中文介绍 作者在 Browserbase 实习期间端到端完成 blogEO 引擎：自动审计、重写和生成 SEO/AEO 定制博客。从提案到落地，展示如何用 AI 驱动内容增长，属工程实战分享。
@VibeMarketer_ · 38.5K 粉丝 · 129.7K 阅 · 504 赞 · 50 转
I am going to show you how to turn every correction, decision, and useful workflow your team gives AI into one company brain that makes everyone better. Because right now, the smartest AI in your
中文介绍 教你把团队给 AI 的每次修正、决策和有用工作流，沉淀为一个不断变聪明的「公司大脑」，让全员受益于集体经验。核心是用系统化复用替代孤立使用 AI，属于企业级 AI 工作流管理方法论。
@EXM7777 · 132.1K 粉丝 · 126.3K 阅 · 504 赞 · 40 转
I'm going to show you how Grok Bot works, then hand you 10 workflows that drive real cash for your business... one specialized bot per workflow, running 24/7 on its own computer and i'll say it
中文介绍 拆解 Grok Bot 的工作原理，并给出 10 个能带来真实收入的工作流：每个工作流配一个专用 bot，在独立电脑上 7×24 小时自动运行。面向想用 AI 实现现金流增长的人，强调直接可落地。
@chamath · 2.4M 粉丝 · 81.1K 阅 · 595 赞 · 52 转
“I actually believe a new asset class will be buying futures of compute. We just don’t have enough compute power right now.” - Larry Fink, CEO of BlackRock Recent news moved that prediction closer to
中文介绍 引用贝莱德 CEO Larry Fink 的判断：未来新资产类别是「算力期货」，因为当前算力供给不足。作者就此展开深度分析，认为近期行业动态正把这一预测推近现实，指向可能的下一个万亿美元市场。
@FareaNFts · 66.2K 粉丝 · 43.7K 阅 · 534 赞 · 39 转
If I wiped Hermes tomorrow and started from zero, I would not open a blank chat and "figure it out." I would install skills first. A raw Hermes install is smart. A skilled Hermes setup is unfair.
中文介绍 如果重置 Hermes，作者会先安装 17 个技能（skills），而不是打开空白对话去「现想」。原始 Hermes 只算聪明，装好技能的 setup 才「不公平」地强大。分享如何用预装技能让 AI 工具 10 倍发挥效能。
@GoogleCloudTech · 1.3M 粉丝 · 35.3K 阅 · 526 赞 · 72 转
How do you ship a software product with 0 lines of manually-written code? A friend asked me this today, and I realized I didn’t have a simple answer. So I dug deeper. It turns out the answer is in how
中文介绍 如何用 0 行手写代码交付软件产品？答案深挖于 harness engineering（驾驭工程）：通过约束与引导 AI 的方式，让想法自动变成产品。Google Cloud 作者解释这一概念为何值得关注。
@_jaydeepkarale · 31.1K 粉丝 · 24.7K 阅 · 506 赞 · 86 转
A phase-by-phase path into AI engineering, with the specific courses, docs, and books worth your time at each stage, not just a list of skills to Google later Most AI engineering roadmaps are skill
中文介绍 分阶段通往 AI 工程就业的路线图：每阶段列出值得投入时间的具体课程、文档和书籍，而非笼统的技能清单。从零基础到可雇佣状态，适合想要系统入行 AI 工程的人。
中文介绍 Hugging Face博客发布文章，介绍IBM Granite 4.2大语言模型是如何构建的，涵盖其设计思路与实现要点。
中文介绍 Hugging Face博客介绍“量化感知修复”技术，该技术能让4位量化压缩模型在性能上超过其全精度原始模型。
Humanoid robots are having a moment in China. The popular machines are part of the country’s strategy to bring artificial intelligence into daily life. Embedding the technology into physical systems—an idea called embodied AI—was a key facet of China’s latest five-year plan, and companies here are a
中文介绍 麻省理工科技评论记者在上海参加了一场机器人“嘉年华”。文章称，人形机器人在中国掀起热潮，这正是国家将AI融入日常生活的战略之一，“具身智能”也是中国最新五年计划的重要方面。
OpenAI CFO Sarah Friar explains how advances across chips, compute, models, and products compound to deliver more useful intelligence at greater scale and lower cost.
中文介绍 OpenAI首席财务官Sarah Friar发文称，芯片、算力、模型与产品的进步相互叠加，正以更大规模和更低成本，交付更有用的智能。
Jalapeño is a custom inference chip from OpenAI that delivers faster, more power-efficient AI inference, with higher throughput and lower latency for modern models.
中文介绍 OpenAI公布自研推理芯片Jalapeño的首批结果，称其可为现代模型提供更快、更节能的AI推理，具备更高吞吐量和更低延迟。
An industry legend starts covering the inevitable!
中文介绍 Latent Space报道，AI行业传奇人物Andrew Ng（吴恩达）进入AI工程领域。文章称，这位行业先驱开始关注这一必然趋势。
OpenAI banned Russia-origin accounts using AI to promote a fake Israel-based think tank and a “sovereignty” index praising Russia and criticizing the West.
中文介绍 OpenAI封禁了一批源自俄罗斯的账户，这些账户利用AI推广一个虚构的以色列智库，以及一个标榜“主权”、吹捧俄罗斯并批评西方的指数。
Use the Admin plugin for ChatGPT Work and Codex to analyze workspace usage, manage members and permissions, adjust limits, and act on admin requests.
中文介绍 OpenAI为ChatGPT Work和Codex推出Admin插件，可分析工作区使用情况、管理成员与权限、调整限制，并处理管理员请求。
中文介绍 Hugging Face博客发布指南，介绍使用Gradio连接、运行和部署AI工作流的方法。
中文介绍 TLDR AI简报（8月25日）关注英伟达Groq芯片、前沿经济学与Ox Alpha之谜。
This article is from Making AI Work, MIT Technology Review’s limited-run newsletter examining how to apply LLMs across industries. To receive it in your inbox, sign up here. Chatbots took many schools by surprise upon their release a few years ago. Suddenly, students carried an app in their phones t
中文介绍 MIT Technology Review文章探讨如何引导学生在课堂上更明智地使用AI。文章称，聊天机器人几年前让许多学校措手不及，学生很快开始使用它们。
GPT‑5.6 is now available in Kiro, helping developers plan, build, review, and test software with better price-performance.
中文介绍 OpenAI宣布GPT-5.6已在Kiro平台上线，帮助开发者规划、构建、审查和测试软件，并提供更优的性价比。
People have been talking to each other for at least 100,000 years, as best we can tell. And in all that time, there has been only one thing in the world that could learn a human language to perfect fluency: a human child. Now there are two. Four short years after the release of ChatGPT,…
中文介绍 MIT Technology Review文章称，儿童学习语言的能力仍优于AI，原因尚不明确。文章指出，史上只有儿童能完美掌握人类语言，而AI发布四年后，现在有了两个。
**Agent harnesses** are becoming a key optimization focus, with NVIDIA research showing traditional skill checks poorly predict agent usefulness and proposing a new metric called **"Skill Lift"**. Open-source implementations of **persistent and self-modifying agents** like **Headlong** and **exo** e
中文介绍 Smol AI News简报称，Agent harness正成为关键优化方向。NVIDIA研究发现传统技能测试难以预测智能体实用性，并提出新指标“Skill Lift”；同时出现了持久性、自修改智能体的开源实现。
中文介绍 TLDR AI简报（8月24日）聚焦DeepSeek Flash Vision、Claude Mythos安全议题，以及Grok Bot内部机制。
美股以中性偏强为主：SPY 贴近 20 日线、RSI 54.2，出现 MACD 死叉但仍处多头排列；QQQ 略低于 50 日线，RSI 48.9；MSFT 单日 +0.9%、RSI 65.7，而 TSLA、META 均呈空头排列。加密市场情绪读数较高，恐慌贪婪指数 65 处于贪婪区间，总市值 2.66T 美元、24h -4.04%；BTC/ETH/SOL 的 RSI 分别为 81.4/76.9/80.7，5 日涨 7.89%/5.59%/10.64%，均显超买。中概分化：BABA 5 日 -6.8%，JD 出现 SMA50 上穿 SMA200 金叉，腾讯控股呈空头排列。商品外汇中黄金 RSI 77 超买、5 日 +5.05%；WTI 原油 5 日 -6.48%；美元兑人民币 RSI 16.7 超卖且空头排列，美元指数 RSI 33.3 偏弱。
当前价 765.91，RSI 54.2 处于中性区间，价格略高于 SMA20(764.8) 和 SMA50(752.75)，低于 52 周高 1.73%。MACD 4.54 低于信号线 5.90，呈死叉，短期动能转弱；整体仍处于多头排列，趋势与短线信号互相抵消，技术状态中性。
当前价 78,796.08，高于 SMA20(68,219.89)、SMA50(65,759.03) 与 SMA200(69,127.04)，5 日上涨 7.89%，MACD 3,746.02 高于信号线 2,167.27，柱体为正，动量偏强；但 RSI 81.4 处于超买区，显示短线上行斜率过大，存在回踩风险。当前技术 setup 偏上行。
当前价 4,716.30，1 日 +1.69%、5 日 +5.05%，价格站上 SMA20(4,370.66)、SMA50(4,203.18) 和 SMA200(4,509.11)，MACD 133.23 高于信号线 97.12，上涨动能维持；RSI 77 已进入超买区，短线过热但未破坏上行结构，技术状态偏上行。
当前价 6.71，RSI 16.7 处于超卖区，价格低于 SMA20(6.74)、SMA50(6.76) 与 SMA200(6.87)，5 日 -0.5%，现价贴近 52 周低点区间；MACD -0.0133 位于信号线下方，继续运行于空头排列。整体技术 setup 偏下行，但超卖状态提示短线存在技术性修复条件。
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
本报告仅基于公开行情数据计算的技术指标进行客观描述，过去走势不代表未来表现，技术信号可能失效，市场受多重因素影响。所有内容仅供技术指标解读参考，不构成任何投资建议。
Families displaced by war in Sudan’s El Obeid refugee camps face crippling water shortages.
中文摘要 苏丹埃尔奥贝德（El Obeid）难民营内的战争流离失所家庭正面临严重水短缺，水资源危机进一步恶化其生存条件，生活状况愈发艰难。
Follow updates live Get our breaking news email, free app or daily news podcast Angela Jackson, the commissioner of the Productivity Commission, said the body stands by the interim report into the GST, which found the deal with Western Australia to be a costly mistake that should be reversed. Jackso
中文摘要 澳大利亚通胀回落不及预期，加息可能性上升；北领地议员卢克·高斯林让位特使。生产力专员安吉拉·杰克逊表示，临时报告认为与西澳的商品及服务税（GST）协议代价高昂且属错误。
Heavy rain expected in parts of Vietnam and southern China, as another storm system, Typhoon Saudel, heads for Japan and Taiwan Widespread flooding triggered by Typhoon Narra has submerged homes and forced mass evacuations in southern China, as another storm system tracked towards Japan and Taiwan.
中文摘要 台风“娜拉”引发创纪录洪水，淹没中国南方房屋并迫使大规模疏散；另一风暴系统“沙特”正向日本和台湾移动，越南及中国华南地区预计有强降雨。
Malaysian leader's comments on the island have prompted discussion about whether the country is leaning towards China.
中文摘要 马来西亚总理安瓦尔对台湾的言论引发讨论，外界关注该国是否正倒向中国，其不结盟立场成为焦点。
Labor put ‘levy on all public transport fares in metropolitan Melbourne and regional Victoria’ since January 2025, auditor-general says The Victorian government secretly increased public transport fares to raise money to help pay for the Suburban Rail Loop and other Big Build projects, the state’s a
中文摘要 澳大利亚维多利亚州政府自2025年1月起秘密上调墨尔本都会区和维州地区的公共交通票价，以筹集资金资助郊区铁路环线（SRL）及其他大型建设项目，审计长证实此事。
Commissioner Angela Jackson also said it was ‘disappointing’ that Anthony Albanese had ruled out changes to Western Australia’s GST arrangements Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast The Productivity Commission has condemne
中文摘要 澳大利亚生产力委员会专员安吉拉·杰克逊表示，西澳州长罗杰·库克称该委员会为“东海岸小丑”已越界；她还对总理安东尼·阿尔巴尼斯排除修改西澳GST安排表示失望。
Army gains, RSF defections and al-Burhan’s political push point to a possible shift in Sudan’s war.
中文摘要 随着苏丹军队取得进展、快速支援部队（RSF）出现叛逃以及布尔汉（al-Burhan）推动政治进程，苏丹战争可能面临转折，战场局势或正塑造下一政治阶段。
New report says abuses against Rohingya and unchecked resource exploitation deepening crisis in Myanmar.
中文摘要 联合国新报告指出，缅甸人权状况“跌至新低”，针对罗兴亚人的虐待和不受约束的资源开采正在加深危机。
Tehran says the agreement with Oman on routes through Hormuz does not indicate that the strait is open.
中文摘要 伊朗方面表示，尽管与阿曼就霍尔木兹海峡航线达成协议，但这并不代表海峡已经开放，该海峡仍处于关闭状态。
Artemis Ghasemzadeh, who had escaped religious persecution, was deported to Panama under Trump’s immigration crackdown. Now, Canada has given her asylum.
中文摘要 阿特米斯·加塞姆扎德因宗教迫害逃离伊朗，在特朗普移民打击下被遣送至巴拿马，如今加拿大已给予其庇护。
The Transportation Safety Board of Canada said the episode involved an Airbus A350 operated by French Bee, a French low-cost airline.
中文摘要 加拿大交通安全委员会称，蒙特利尔机场一名工人在与飞机相关的事故中死亡，涉事飞机为法国低成本航空公司French Bee运营的空客A350。
The United States threatened sanctions for any country or entity that engages with Iran’s gold, digital assets, aviation, shipping and tech industries. Here’s why that matters.
中文摘要 美国威胁对任何与伊朗黄金、数字资产、航空、航运和科技行业往来的国家或实体实施制裁，此举被视为“经济D-Day”，影响远超石油领域。
Owner ByteDance ordered to erase illegally obtained child data in Brazil as crackdown on tech giants intensifies.
中文摘要 巴西因TikTok违反儿童数据隐私规定对其处以3000万美元罚款，并要求母公司字节跳动删除非法获取的儿童数据，这是巴西加强对科技巨头监管的举措。
Why can’t America agree on what time it is?
中文摘要 美国国内对时间标准（包括夏令时制度）长期存在争议，各州和联邦层面难以统一立场，导致全国无法就“现在几点”达成共识。
The amended law was an attempt by state Republicans to crack down on unproven claims of voting by noncitizens.
中文摘要 美国一名联邦法官阻止俄亥俄州实施要求选民登记时提供公民身份证明的法律，该修正案是州共和党人为打击非公民投票的未经证实指控而推动的。
It will be the firm's biggest site yet and aims to send rockets to space in 2029.
Seoul hopes polar sea highway will reduce reliance on chokepoints such as Hormuz and revive ailing city of Busan
Australia’s second-largest pension has built its biggest overweight position in the Japanese yen in years, betting that markets are underpricing Bank of Japan interest rate hikes.
Citadel Securities’ Frank Flight, who just last month warned of a “cruel summer” for US bond investors, now sees the balance of risks tilted toward a rally, citing crowded bearish positioning and improving inflation data.
Xi Jinping’s government is sending a message of defiance to Donald Trump over China’s economic ties with Iran. Bloomberg's Michael Heath breaks down the options available to Donald Trump to strangle Iran's economy and end a conflict that's pulling US resources from Asia. (Source: Bloomberg)
Chinese lenders have started pricing bonds off the nation’s overnight funding cost, an indication that it’s increasingly the benchmark for the financial system.
Firm one of several professional services businesses targeted by the American Alliance for Equal Rights
Gold consolidated after a five-day rally, with investors turning their focus to the Federal Reserve’s interest-rate path ahead of the annual Jackson Hole gathering this week.
Bearish bets against shares of Z.AI Co. and MiniMax Group Inc. have soared to records ahead of their earnings reports, highlighting growing investor worry over China’s heated AI rivalry.
The price of Bitcoin has risen in August but is still far below its peak almost a year ago - and companies are refitting their mines.
After last year's massive power failure Spanish and Portuguese firms are buying battery back ups.
From baby banks to supermarket deals, there are many ways parents can ease the back-to-school burden.
6 回复 · Apple 节点
9 回复 · Apple 节点
5 回复 · 程序员 节点
6 回复 · 程序员 节点
10 回复 · 程序员 节点
7 回复 · Apple 节点
15 回复 · 程序员 节点
9 回复 · Apple 节点
8 回复 · 程序员 节点
26 回复 · Apple 节点
该源今日无内容。