JavaScript · ★ 14,850 · 🍴 813 · 📈 3,006 stars today
A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings
中文介绍 面向 AI 编码助手的网络安全审计技能插件，支持多阶段审计流程，输出可机器读取且经过独立验证的漏洞发现，适合集成到自动化安全巡检、CI 检测与安全评审流程中。
HTML · ★ 23,836 · 🍴 1,648 · 📈 383 stars today
Scale computer-use 2.0 with open-source drivers, cross-OS fleets, and benchmarks for training, evaluation, and data generation.
中文介绍 面向 Computer Use 2.0 的开源基础设施，提供跨操作系统驱动、多设备集群与基准测试能力，覆盖训练、评估和数据生成全流程，服务于 Agent 研究者与大规模桌面/移动端自动化团队。
JavaScript · ★ 96,637 · 🍴 10,209 · 📈 675 stars today
Production-grade engineering skills for AI coding agents.
中文介绍 为 AI 编码代理提供生产级工程能力集，涵盖代码重构、调试、性能优化等真实工程场景。适合将 LLM Agent 接入实际开发工作流的团队与开发者参考使用。
Go · ★ 15,430 · 🍴 1,509 · 📈 478 stars today
Secure environments for developers and their agents
中文介绍 为开发者及其 AI Agent 提供安全隔离的开发环境，支持云端沙箱、资源管控与团队协作。适合需要托管 Agent 执行、远程开发和自动化流水线的工程组织。
TypeScript · ★ 146,492 · 🍴 23,831 · 📈 444 stars today
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
中文介绍 运行在终端中的 Agentic 编程工具，能理解代码库、执行常规任务、解释复杂代码并处理 Git 工作流，帮助开发者在本地环境中更高效地编写与迭代代码。
TypeScript · ★ 15,672 · 🍴 2,052 · 📈 477 stars today
OpenStock is an open-source alternative to expensive market platforms. Track real-time prices, set personalized alerts, and explore detailed company insights — built openly, for everyone, forever free.
中文介绍 开源的股票行情平台，提供实时价格、个性化预警与公司详情分析，定位昂贵的商业行情软件的免费替代。适合个人投资者、量化研究者与教学演示场景。
Jupyter Notebook · ★ 4,652 · 🍴 868 · 📈 325 stars today
Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters
中文介绍 故障容忍且高可扩展的 GPU 编排与机器学习框架，专为十亿至万亿参数规模的大模型训练设计。适合大规模分布式训练、超大规模 AI 基础设施与高性能计算场景。
Python · ★ 66,712 · 🍴 4,815 · 📈 94 stars today
Get your documents ready for gen AI
中文介绍 面向生成式 AI 的文档预处理工具，将 PDF、Office 等复杂文档解析为结构化表示，便于 RAG、检索与 LLM 输入。适用于知识库构建、企业文档问答与数据清洗管线。
Rust · ★ 11,851 · 🍴 1,098 · 📈 5 stars today
🥧 Savoury implementation of the QUIC transport protocol and HTTP/3
中文介绍 Rust 实现的 QUIC 传输协议与 HTTP/3 栈，提供高性能、生产级的网络协议能力。适合构建低延迟网络服务、CDN、实时应用与下一代互联网基础设施。
Go · ★ 5,077 · 🍴 212 · 📈 889 stars today
Your own search engine
中文介绍 可自托管的本地搜索引擎项目，允许用户在本地机器上索引与检索内容，保障数据隐私。适合注重隐私的开发者、家庭服务器用户与小团队内部知识检索。
★ 102,946 · 🍴 4,420 · 📈 144 stars today
科技爱好者周刊，每周五发布
中文介绍 阮一峰主编的科技爱好者周刊，每周五发布，收录前沿技术、开源项目与行业动态。适合作为程序员与科技从业者的轻量阅读与信息聚合入口。
TypeScript · ★ 37,651 · 🍴 5,198 · 📈 31 stars today
人人都能用英语
中文介绍 面向普通人的实用英语学习资料集合，提供基础口语、日常表达与学习方法。适合英语初学者、非母语职场人士自学与日常对话场景。
Python · ★ 24,972 · 🍴 2,978 · 📈 299 stars today
Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork
中文介绍 为知识工作者在 Claude Cowork 中使用而设计的开源插件仓库，覆盖文档处理、信息整理等知识型任务。适合内容创作者、研究人员与企业知识管理场景。
Python · ★ 11,392 · 🍴 727 · 📈 207 stars today
Automation foundation model for tiny devices: 2-bit, 8-29 MB, tool calls, structured extraction and embeddings on phones, wearables, smart homes, robots, cars and microcontrollers.
中文介绍 面向小型设备的自动化基础模型，2-bit 精度、体积 8–29MB，支持工具调用、结构化抽取与嵌入。适用于手机、可穿戴、智能家居、机器人与微控制器等端侧 AI 场景。
Rust · ★ 3,303 · 🍴 440 · 📈 64 stars today
OpenAI Codex 桌面端/CLI 的可视化管理工具，具有Provider/API 切换、会话同步、提示词注入、Skills/MCP 管理、TOML 配置可视化的跨平台工具。
中文介绍 OpenAI Codex 桌面端与 CLI 的可视化管理工具，支持 Provider/API 切换、会话同步、提示词注入、Skills/MCP 管理与 TOML 配置编辑，跨平台运行。
👍 90
Recent Omni-Modal Generative Models (Omni-Models) have advanced content generation toward unified modeling of text, images, video, and audio. MiniMax-H3 exemplifies this transition by combining multimodal context understanding with joint audio-visual generation in a shared latent framework. Its unif
中文介绍 研究 Omni-Modal 生成模型 MiniMax-H3 在物理世界推理上的能力。该模型在共享潜空间中联合建模文本、图像、视频与音频，并进行多模态上下文理解与音画联合生成。文章系统评测其物理规律遵循与场景一致性表现，指出当前通用生成模型在物理因果推理上仍存在不足。
👍 82
The widespread adoption of long-horizon agents has made model workloads increasingly input-heavy. Although prior work has substantially reduced the cost of long-context computation, prefill remains computationally expensive, and large KV caches continue to strain HBM and SSD capacity and data-transf
中文介绍 针对长程 Agent 场景下 prefill 开销大、KV cache 占用 HBM/SSD 及带宽的问题，提出 DeepSeek-V4.1-Flash 的 KV cache 压缩方案。通过更激进的压缩与重构策略降低长上下文推理的存储与传输成本，为超长输入推理提供更高效部署基础。
👍 73
Scientific code repositories encode decades of human knowledge in executable models, methods, and tools. Yet fragmented toolchains, implicit domain conventions, and specialized correctness criteria make this knowledge difficult to convert into reliable learning experience-a challenge we call the sci
中文介绍 提出 ScienceIDE，将世界科学代码库转化为可供 Agent 学习的可执行环境。针对工具链碎片化、领域约定隐式化与正确性标准专门化所导致的知识转化难题，构建包含可运行代码、验证标准与领域知识的环境，使 Agent 获得可靠的科学计算学习体验。
👍 70
We study length inflation in on-policy distillation (OPD), where student responses can become excessively long and even exhaust the generation budget. We identify termination-token mismatch between base students and post-trained teachers as an important source of this behavior. Across Qwen3, Llama,
中文介绍 研究 on-policy distillation (OPD) 中学生回复过长乃至耗尽生成预算的 length inflation 问题。识别出 base student 与 post-trained teacher 之间 EOS/termination token 不一致是关键诱因，在 Qwen3、Llama 等模型族上分析并讨论缓解该长度膨胀的机制与做法。
👍 65
As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated predictions into long trajectories of reasoning, tool use, and feedback. Token efficiency therefore becomes important for scaling recursive self-improvement. We take an
中文介绍 提出 SoL-Pi，将自动化研究循环递归扩展用于构建高效的 Agent harness。面向无人值守的长轨迹代码探索，关注推理、工具调用与反馈中的 token 效率，以提升递归自改进的扩展性，使 Agent 在更长的探索轨迹中保持成本可控。
👍 54
Reliable confidence estimation is increasingly central to the trustworthy deployment of language models: a calibrated estimate of the probability that an output is correct decides what to ship, what to escalate, and what to retry. Existing confidence estimators, however, share one design premise: th
中文介绍 提出从推理到 Agent 的经验式置信度估计方法。针对现有置信度估计器共享同一设计前提而难以适应复杂部署的问题，利用 Agent 积累的经验对输出正确概率进行校准，用于决策发布、升级或重试，提升模型部署的可信度。
👍 52
Coding harnesses shape how autonomous coding agents translate model capabilities into long-horizon software-engineering performance, yet existing work typically evaluates harnesses as monolithic systems, leaving the effectiveness of individual components unclear. To enable component-level comparison
中文介绍 对 coding agent 的 harness 设计进行实证研究。针对现有工作常把 harness 当作整体系统评估、难以厘清各组件有效性的问题，设计可组件级对比的评测框架，拆解分析各 harness 要素如何影响长程软件工程表现，为设计选择提供依据。
👍 46
Coding agents are typically evaluated with desired behavior specified through issues or instructions. In practical web development, however, agents may need to infer behavior from working software and implement it in an incomplete application. We introduce ProgramDistill, a benchmark evaluating codi
中文介绍 提出 ProgramDistill 基准，将交互式 Web 应用转化为可验证的、参考引导的 SWE 任务。区别于仅以 issue 或指令描述期望行为的评测方式，要求 Agent 从可运行的软件中推断行为并在不完整应用中复现，更贴近真实 Web 开发需求。
👍 41
Autonomous research loops such as AutoResearch show that one coding agent can improve a training setup unattended. Run several of them and each session starts from scratch, so more agents tend to mean more duplicated search rather than more discovery. Agora is a shared memory for such agents: resear
中文介绍 提出 Agora，将 Git 作为共享内存用于集体 AutoResearch。针对多 Agent 并行运行导致重复搜索而非更多发现、各会话从零开始的问题，Agora 让 Agent 在共享 Git 上下文中协作、复用探索成果，提升自主研究循环的整体效率。
👍 36
World modeling enables intelligence to anticipate consequences, guide interventions, and learn from interaction. Yet predictive models remain domain-specific: can a common learning principle support world modeling across radically different systems? We introduce JEPA-Anything, a domain-agnostic fram
👍 31
Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signal
中文介绍 提出 EvolveTrade，用经验驱动策略精细化让 LLM 交易 Agent 自我演化。针对静态手写工具调用策略在部署前固定、难以适应证据收集、工具调用与信号验证方式的问题，依据交易经验动态调整策略，提升 Agent 适应能力。
👍 29
Platform abuse campaigns conceal redirection instructions with emojis, homophones, character decomposition, and redundant symbols, then route users through disguised links to services associated with pornography, fraud, gambling, or illicit transactions. Existing benchmarks evaluate obfuscated text
中文介绍 提出 RiskChainBench，评测混淆平台消息还原与基于证据的 Web 调查能力。针对滥用活动用 emoji、同音字、字符拆分、冗余符号隐藏重定向并诱导用户进入色情、诈骗、赌博或非法交易服务的问题，评测从还原文本到追踪关联证据的能力。
👍 27
LLM assistants are widely used for daily social advice, yet evaluating their social reasoning in such consultation settings remains challenging since (i) it requires setups where the assistant learns about social situations from subjective user narratives, and (ii) social properties, such as others'
中文介绍 提出面向 LLM 助手的可验证社交推理评测方法。针对日常社交建议场景中社交推理难以评估的问题，设置 Assistant 从主观用户叙述中学习社交情境、并验证关于他人动机与意图等主观属性的推理过程，提升社交建议的可验证性。
👍 26
Multi-turn agents trained with reinforcement learning (RL) receive a single scalar reward per trajectory, which motivates self on-policy distillation (OPD) to supply dense token-level supervision from a self-teacher with privileged task skills, letting a skill-free student internalize them. This rec
中文介绍 提出 RetireOPD，用于 agentic 强化学习的自淘汰 on-policy distillation。针对多轮 Agent 用 RL 训练时每轨迹仅有单一标量奖励的问题，借助拥有特权技能的 self-teacher 提供密集 token 级监督，并设计机制让 student 在掌握技能后主动淘汰该蒸馏信号。
👍 25
Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations. Recent systems combine scientific agents with evolutionary search through critique, comparison, and revision. However, how different forms of agent collaboration affec
中文介绍 提出 HypoEvolve，用遗传算法使多 Agent LLM 发现科学假设。针对现有系统将科学 Agent 与演化搜索结合但不同协作方式影响不清晰的问题，研究通过 critique、比较与修订等不同 Agent 协作形式如何影响假设发现质量。
👍 24
GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate plans fixed before execution. Recent agent-skill frameworks encapsulate reusable procedural knowledge to mitigate this, yet existing skill designs are
中文介绍 提出 Reflect, Revise, Reuse，一种免训练的 GUI Agent 技能演化方法。针对弹窗、延迟加载、控件位移等动态界面使执行前固定计划失效的问题，让 Agent 通过反思、修订与复用演化可重用过程性知识，无需额外训练即提升长程任务适应力。
👍 22
Document parsing converts document images into structured content and requires reliable performance across diverse layouts and acquisition conditions. Yet training corpora are biased toward common document types and clean digital pages, while expanding coverage alone does not specify how to address
中文介绍 提出 WeVisDoc，从覆盖度转向能力以构建稳健的端到端文档解析。针对训练语料偏向常见文档类型与干净数字页、且单纯扩大覆盖无法说明如何针对能力加以解决的问题，从能力维度设计解析方案，提升跨版式与拍摄条件下的鲁棒性。
👍 22
Direct preference alignment methods are widely used to align large language models (LLMs) with human preferences because of their computational and memory efficiency. However, likelihood displacement motivates alternative ways to extract information from preference pairs with small likelihood margin
中文介绍 提出用于 LLM 偏好对齐的零阶优化范式。针对直接偏好对齐方法虽在算力与显存上高效，却因 likelihood displacement 难以从小似然边距偏好对中提取信息的问题，提出以零阶方式替代似然梯度以提取偏好信号进行对齐。
👍 19
Information retrieval is increasingly important as LLM agents tackle complex tasks involving diverse information needs. Because retrieval relies on an index that represents each document through index keys, retrieval quality depends heavily on how effectively these keys expose the knowledge containe
中文介绍 提出 Self-Evolving Search Index，面向 LLM Agent 的自演化检索索引。针对复杂任务中检索质量严重依赖 index key 能否有效暴露文档所含知识的问题，使索引能够随使用自演化、动态优化 key 表示，从而提升多样化信息需求下的检索效果。
👍 19
Spatial intelligence requires more than describing object locations. Under incomplete observation, models must identify and acquire missing evidence, interpret it in a common spatial frame, and act on it. We introduce VA-Bench to evaluate the complete observe-reason-act-revise loop. General-purpose
中文介绍 提出 VA-Bench，通过视觉演示、主动感知与度量控制评测具身空间智能。针对空间智能不仅是描述物体位置、还要求模型在不完整观测下主动获取缺失证据、在统一空间框架内解释并据此行动的问题，评测完整的 observe-reason-act-revise 闭环。
@sandylikesfrogs · 3.9K 粉丝 · 367.0K 阅 · 512 赞 · 36 转
What we learned making our first $75 with an AI-run business. In July, we decided to see if an AI agent could run a company that could actually make money. We gave Devin a Ramp card, a phone number
中文介绍 团队尝试让AI代理自主经营赚钱：给Devin配置Ramp卡、手机号等资源，总结首个75美元收入的实战经验与踩坑。
@sairahul1 · 145.3K 粉丝 · 330.3K 阅 · 568 赞 · 59 转
You do not need a 70B model. You do not need $100,000 in compute. You do not need a machine learning team. A 1.5B parameter model fine-tuned on 200-500 good examples can outperform a frontier model
中文介绍 LLM微调入门教程：用200-500个优质样本微调1.5B小模型即可胜过前沿大模型，无需70B参数、10万美元算力或专业团队。
@mustafasuleyman · 1.1M 粉丝 · 204.3K 阅 · 502 赞 · 86 转
AIs are not conscious. They do not feel, experience, or suffer. However, there’s a growing movement in support of model welfare; the idea that we might soon owe them a duty of care. I think this
中文介绍 讨论「模型福利（model welfare）」议题：AI无意识但业界兴起对模型负关怀义务的运动，作者对此观点提出探讨。
@Sabrina_Ramonov · 7.1K 粉丝 · 160.5K 阅 · 518 赞 · 398 转
I talk to business owners who have piles of photos and videos, but no time to post to social media. They don't know what caption to write. They don't know how to edit videos. So, all that good
中文介绍 面向新手的实操教程：用GPT-6 Astra结合DaVinci MCP，帮店主处理堆积的照片视频，自动写文案、剪辑并发布社媒。
@chddaniel · 27.7K 粉丝 · 121.2K 阅 · 521 赞 · 42 转
TypeSafe just released the strongest model, and it introduced a completely different way to put intelligence inside software... Jev is in a category of its own right now, not because it writes better
中文介绍 Jev入门全指南：TypeSafe推出最强模型Jev，主打把决策智能嵌入软件，不是靠写作更好，而是开辟全新类别。
@mvanhorn · 40.4K 粉丝 · 103.6K 阅 · 550 赞 · 37 转
Here is the tl;dr ELI5. Think AI multiple choice, not AI essay writing. It doesn't chat. It makes decisions your software can act on: "Spam or not?" "Which tool should this agent use?" "Does this need
中文介绍 Jev科普加案例：把它理解为AI选择题而非写文章，用于让软件做决策（如判断垃圾邮件、选工具），并列出9个已落地应用。
@sydneyrunkle · 10.6K 粉丝 · 88.0K 阅 · 556 赞 · 66 转
Agents run in a loop: an LLM decides what to do, a tool executes, a model evaluates the results, and then continues in that loop until the task is complete. Agents and LLMs were initially difficult to
中文介绍 用Jev搭建Agent Harness：解释Agent循环（LLM决策→工具执行→模型评估→继续）的工作机制，并演示如何落地搭建。
@JacquelineSYC19 · 1.0K 粉丝 · 72.4K 阅 · 501 赞 · 47 转
At the time of writing, Artie is a team of 17. Every one of us works alongside Hermes, an open-source AI agent harness from @NousResearch. It runs on a single physical machine in Germany, has a
中文介绍 团队案例：Artie公司17人全部与开源AI代理Hermes协作，部署在德国一台物理机上，展示如何支撑整个团队日常工作。
@0xCodila · 18.6K 粉丝 · 63.4K 阅 · 521 赞 · 62 转
The Jevons Paradox (he's on picture) is a rule stating that an increase in the efficiency of a resource's use does not reduce, but rather increases, its overall consumption - That's the global
中文介绍 Jev Engineering十步路线图：从原理（引用杰文斯悖论：效率提升反而增加资源消耗）到从零搭建AI新大脑的完整设置流程。
@akshay_pachaar · 286.8K 粉丝 · 41.6K 阅 · 578 赞 · 87 转
We have been using LLMs like a hammer for every AI problem, even simple decisions. Jev handles those decisions in milliseconds at a fraction of the cost. Let's understand how it works and where it
@ShaneLegg · 82.3K 粉丝 · 40.6K 阅 · 585 赞 · 86 转
We are on the cusp of a profound transformation. Today’s AI systems have impressive capabilities and the rapid pace of innovation suggests we’re now approaching artificial general intelligence (AGI),
中文介绍 Shane Legg宣布成立DeepMind Institute，提出当前AI能力快速逼近AGI，呼吁从研究转向应对即将到来的深刻转型。
@Didotxyz_ · 64.6K 粉丝 · 5.7K 阅 · 502 赞 · 325 转
The important shift in AI agents is not the disappearance of chat. It is that chat is no longer the whole product. An assistant explains a task. An agent system has to carry it forward: call a tool,
中文介绍 论述AI代理趋势：不是对话消失，而是对话不再是产品全部——助手只解释任务，代理系统要真正调用工具把任务完成。
Imitation is the sincerest form of Flattery
中文介绍 Latent Space的AINews报道，有人在两天内完成了对AI工具Jev的6个仿制品，反映了当前AI生态中快速模仿与创新的现象。
OpenAI introduces the Australian Youth Safety Blueprint, a six-pillar roadmap for safer AI experiences that protect and empower young people.
中文介绍 OpenAI发布《澳大利亚青少年安全蓝图》，提出六支柱路线图，旨在为青少年提供更安全的AI体验，兼顾保护与赋能。
On Wednesday, MIT Technology Review hosted a live Roundtables event for subscribers that asked the question everyone’s asking right now: Could AI really kill us all? But attendees had so many more questions than we had time to answer in the 30 minute session. So we asked our senior AI editor Will Do
中文介绍 MIT Technology Review举办线上圆桌讨论，回答'AI是否真的能杀死所有人'等公众关切，编辑整理了未在30分钟内解答的更多问题。
a quiet day
中文介绍 Latent Space的AINews指出当日AI领域并无重大新闻，为相对平静的一天。
中文介绍 TLDR AI日报提及Claude Projects v2发布，Google推出家庭智能代理等AI行业动态。
Cooley built GO Public with ChatGPT Work to bring intelligence to the IPO process, helping lawyers surface issues earlier and focus judgment where it matters most.
中文介绍 Cooley律师事务所基于ChatGPT Work构建GO Public工具，将智能引入IPO流程，帮助律师更早发现问题，将判断力集中在最关键环节。
A dash of cold water keeps the foomers away.
中文介绍 Latent Space的AINews对AI新闻进行冷静审视，包括Yegge关停Gas Town项目、Databricks的Astra成本上升60%等内容，以理性视角抑制过度炒作。
OpenAI for Law brings frontier intelligence for law, custom firm workflows, connected legal data sources, and legal-grade controls for confidential client work.
中文介绍 OpenAI推出面向法律行业的Astra for Law产品，提供前沿AI智能、自定义律所工作流、关联法律数据源及法律级保密控制能力。
中文介绍 TLDR AI日报报道Claude与Cowork合并、ChatGPT推出赞助代理及税务harness等AI行业最新动态。
We sit down with AIUC’s CEO on their Series A!
中文介绍 Latent Space报道AIUC首席执行官Rune Kvist专访，该公司完成A轮融资，主打可被追责的AI Agent，推动智能监管体系建设。
OpenAI shares a framework for tracking, investigating, and disclosing model misalignment, alongside six reports of unexpected or concerning model behavior.
中文介绍 OpenAI发布模型错位报告框架，用于跟踪、调查和披露模型偏差，并公布了六份关于模型出现意外或令人担忧行为的报告。
OpenAI and AARP are bringing free, hands-on ChatGPT workshops to 1,000 older adults across 10 U.S. cities to build practical AI skills safely.
中文介绍 OpenAI与AARP合作，在美国10座城市为1000名老年人提供免费ChatGPT实操工作坊，帮助其安全掌握实用AI技能。
Explore new AI-powered advertising experiences from OpenAI, including Sponsored Agents, tools for marketers, and integrations with HubSpot and Shopify.
中文介绍 OpenAI推出AI驱动的广告新体验，包括赞助代理、营销工具和与HubSpot、Shopify的集成方案。
The AI boom is becoming a materials challenge. As AI pushes computing into new territory, the materials behind that infrastructure are becoming just as crucial as the algorithms running on it. Semiconductors and data centers are approaching physical limits around performance, thermal management, ele
中文介绍 MIT Technology Review报道AI热潮正演变为材料挑战，半导体和数据中心在性能方面逼近物理极限，基础设施材料变得与算法同等关键。
GPT-6 Astra helps Hex’s data agents turn answers into interactive visualizations that employees are proud to share.
中文介绍 Hex利用GPT-6 Astra将复杂数据分析转化为可视化报告，帮助员工生成可分享的交互式图表，提升数据协作效率。
美股指数层面 SPY 报价 761.69、较 52 周高仅 -2.27% 且处于 SMA200 上方 6.4%，RSI14=49.2 中性，MACD 短线由正转负但仍维持多头排列；QQQ 报价 721.45、近 5 日 +0.92%、MACD 金叉（MACD 0.2176 > Signal 0.1465）、RSI14=55.9，科技股动能略占优。加密侧恐慌贪婪指数 71（贪婪），总市值 2.79T USD、24h +0.9%，BTC 主导率 58.3%、ETH 11.5%；BTC-USD 81253.62、5 日 +3.95%、多头排列但距 52 周高 -35.61% 仍深跌修复中；ETH-USD 2641.36、5 日 +5.05%，SOL-USD 5 日 +9.15% 且触发 MACD 金叉，加密短周期动量偏强。中概股集体偏弱：BABA、PDD、0700.HK 均空头排列，BABA 虽 1 日 +4.33% 仍距 52 周高 -41.23%、趋势 bearish；JD 报价 26.92、RSI14=34.8 接近超卖。商品外汇端 CL=F 96.08 出现 1 日 -5.72% 的动能回落但 RSI14=63.8 仍偏强；GC=F 4424.9 处于 SMA50 上方但 MACD 绿柱，RSI14=50.3 中性；USDCNY=X 6.70、RSI14=21.9 超卖、空头排列，接近 52 周低。宏观 ^VIX 14.81、5 日 -6.5% 显示避险情绪降温，而 ^TNX 5.00% 多头排列、RSI14=68.8 偏强，DX-Y.NYB 100.22 多头排列且接近 52 周高，美元强、美债收益率上行的组合对中概与商品端形成压制。
报价 111.87、近 1 日 -0.65% 但近 5 日 +9.15% 动量最强，SMA20=102.70、SMA50=91.50、SMA200=83.67 完整多头排列，MACD 3.5532 > Signal 3.3976 触发金叉，RSI14=65.1 偏强未超买，趋势 bullish，距 52 周高仍 -53.74% 具备修复空间。
报价 721.45、近 1 日 +0.63%、近 5 日 +0.92% 表现强于 SPY，SMA20=713.24 < 现价、SMA50=709.95、SMA200=662.60 多头排列，MACD 0.2176 上穿 Signal 0.1465 形成金叉，RSI14=55.9 中性偏强，距 52 周高 -3.63% 接近突破区。
报价 336.13、近 5 日 +1.16%、距 52 周高仅 -2.45% 处于突破前夜，SMA20=322.64、SMA50=320.14、SMA200=286.34 完整多头排列，MACD 5.5173 大幅高于 Signal 3.9146 红柱扩张，RSI14=64.3 偏强但未进入超买，52 周低 +39.93% 趋势强度突出。
报价 419、近 1 日 -1.64%、近 5 日 -2.19% 持续走弱，SMA20=437.42、SMA50=452.50、SMA200=506.86 三层均线全在上方的空头排列，MACD -7.3325 低于 Signal -6.4129 绿柱仍在放大，RSI14=36.9 偏弱，信号标记「接近 52 周低」距 52 周高 -38.65%。
报价 6.70、近 5 日 -0.16%、距 52 周高 -6.86% 但仅高于 52 周低 0.14%，SMA20=6.71、SMA50=6.74、SMA200=6.84 呈空头排列，MACD -0.0100 仍低于 Signal -0.0103 且方向向下，RSI14=21.9 已进入超卖区，trend=bearish，信号含「RSI 超卖」「接近 52 周低」「空头排列」。
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
以上分析仅基于公开行情计算的技术指标读数，过去走势不代表未来表现，本内容仅供技术指标解读参考，不构成任何投资建议或买卖指引，市场风险由使用者自行承担。
Widespread power outages have become increasingly common due to fuel shortages made worse by a US blockade.
中文摘要 古巴因燃料短缺再遭大范围停电，数百万人无电可用。BBC报道称，美国封锁加剧了燃料危机，导致大规模停电日益频发。
Many streets in downtown Montreal and along the south shore of the St. Lawrence River will become a giant sports arena for 1,000 cyclists.
中文摘要 半个世纪后，公路自行车世锦赛重返加拿大蒙特利尔。市中心及圣劳伦斯河南岸多街道将封闭，供约1000名车手参赛，形成大型赛事场地。
New agreement with Denmark falls short of Trump’s previously stated goal of US 'ownership' of the island.
中文摘要 特朗普称美国对格陵兰安全拥有「永久控制权」，但新达成的美丹协议未实现其此前所提美国「拥有」该岛的目标。半岛电视台质疑其法律实际效力。
The promotion of monocultures is hurting the production of much more nutritious and climate-resistant traditional crops.
中文摘要 非洲绿色革命推广单一作物种植，挤压了营养更丰富、抗气候能力更强的传统作物生产，威胁当地传统食品体系。
President Trump signed the legislation, named for Senator Lindsey Graham, at the White House, but he has broad leeway to waive its requirements.
中文摘要 特朗普在白宫签署以参议员林赛·格雷厄姆命名的对俄新制裁法案。俄方警告该法案将损害特朗普的和平努力，但特朗普可广泛豁免其要求。
Police boats and helicopters are patrolling Perth's beaches as authorities continue their search for the swimmer's body, and the shark that attacked him.
中文摘要 西澳大利亚发生致命鲨鱼袭击，当局罕见下令猎杀涉事鲨鱼。珀斯海滩部署警船和直升机巡逻，继续搜寻失踪游泳者遗体及该鲨鱼。
美国总统特朗普9月18日宣布，已经和丹麦、格陵兰方面达成了一项协议，美国将获得格陵兰防务的永久控制权。丹麦和格陵兰方面也对此表示欢迎，强调格陵兰会变得更安全，从而也让欧洲国家更安全。
中文摘要 特朗普9月18日宣布与丹麦、格陵兰达成协议，美国将获格陵兰防务永久控制权。丹麦和格陵兰方面表示欢迎，称格陵兰及欧洲国家将更安全。
根据法案，美国总统还可以对在帮助俄罗斯规避能源制裁方面出力最多的五个国家实施制裁。这些措施可能进一步加剧美国与中国和印度的贸易关系紧张，因为两国都大量购买俄罗斯石油。
中文摘要 特朗普签署对俄伊新制裁法案，可对协助俄罗斯规避能源制裁的五个国家实施制裁。此举或加剧美国与中国、印度贸易关系紧张，两国大量购买俄石油。
The United Nations General Assembly will allow Palestinian President Mahmoud Abbas to address via a prerecorded video.
中文摘要 欧盟敦促美国解除对巴勒斯坦代表团的旅行禁令，以使其出席联大会议。联合国大会将允许巴勒斯坦总统阿巴斯通过预录视频发表讲话。
In the eastern city of Xalapa, what started as a joke among friends over two decades ago has become a vibrant annual tradition that highlights creativity and absurdity with melons.
中文摘要 墨西哥东部城市萨拉普阿举办一年一度的「最美西瓜」选美比赛，源于二十多年前朋友间的玩笑，如今已成为突出创造力与荒诞趣味的传统活动。
Jailed in Madrid on an extradition request from the United States that his lawyer says is tied to pro-Palestinian activism, James Chambers fears persecution if returned.
中文摘要 百万富翁活动家詹姆斯·钱伯斯在马德里被依据美方引渡请求拘留，其律师称引渡与亲巴勒斯坦活动有关。他向西班牙申请政治庇护，担心被遣返后遭迫害。
President Vladimir V. Putin of Russia has said veterans are the “new elite,” but selecting them as candidates is a risk.
中文摘要 普京称退伍军人是「新精英」，克宫将其大量推入选举候选名单。纽约时报指出此举存在反噬风险，或影响选举结果。
Exxon Mobil found a coveted growth engine in Guyana. Now it must navigate Guyana’s fractious politics and a territorial dispute with Venezuela.
中文摘要 美国最大石油公司埃克森美孚在圭亚那找到新的增长引擎，但需应对圭亚那复杂的政治局势以及与委内瑞拉的领土争端。
Oil-rich Guyana is enjoying red-hot growth. But the boom is magnifying tensions over corruption, inequality and strained ties with Venezuela.
中文摘要 受伊朗战争影响，富油小国圭亚那迎来高速增长，但繁荣同时放大了腐败、不平等以及与委内瑞拉关系紧张等矛盾。
The ban is the latest attack by Trump on media outlets that he perceives as unfavourable to his administration.
中文摘要 特朗普能否禁止CNN等其不满的新闻媒体进入白宫引发讨论。半岛电视台称，此举是特朗普针对其认为对政府不利的媒体的又一次攻击。
As Europe heads into winter, electricity markets are flashing their strongest warning since the energy crisis.
中文摘要 彭博报道，欧洲进入冬季前，电力市场发出自能源危机以来最强烈的预警信号，电价大幅上涨，供应紧张风险上升。
Gene Sykes on the TMT bubble and more.
中文摘要 高盛并购银行业务负责人Gene Sykes在接受采访时，谈到了自己如何帮助促成2028年洛杉矶奥运会的落地，并回顾了TMT泡沫时期的经历。
Also in today’s newsletter: Clearlake wins the Chelsea civil war
中文摘要 英国《金融时报》报道，某俱乐部通过签下姆巴佩成功打击陷入困境的耐克，同时Clearlake在切尔西内部博弈中胜出。
We’re in an era of megadeals, with Goldman Sachs reporting a record year for big deal volumes as companies scramble to compete with AI. Gene Sykes has seen multiple deal cycles come and go over his multi-decade career. The co-head of mergers and acquisitions has worked on famous transactions like Co
中文摘要 彭博《Odd Lots》播客中，高盛并购联合主管Gene Sykes回顾多年并购周期，谈到公司在AI竞赛推动下录得创纪录的大额交易年份，并分享参与2028奥运会交易的经历。
Bolivia’s government has completely eliminated its diesel subsidy after the Senate approved a $1.9 billion loan from the International Monetary Fund that calls for spending curbs.
中文摘要 玻利维亚参议院批准国际货币基金组织19亿美元的贷款后，政府完全取消了柴油补贴，贷款条件要求实施支出削减。
Growing interest comes amid a rise in larger annuity purchases
中文摘要 英国《金融时报》报道，随着利率上升，80岁以上人群对年金产品的兴趣日益增长，大额年金购买量也在增加。
Not even I am immune from this late-night drain on my bank account
中文摘要 英国《金融时报》专栏文章讨论个人财务管理中最糟糕的习惯，作者坦言自己也难以避免深夜消费对银行账户的侵蚀。
Report finds some chatbots ignored upcoming tax changes and hallucinated rules
中文摘要 英国《金融时报》报道，报告显示AI聊天机器人在回答财务问题时「大多数时候」给出错误答案，部分模型忽略即将到期的税务变更并凭空捏造规则。
Rents and tenant demand are rising, but landlords are under increasing pressure.
中文摘要 英国《金融时报》回顾买租物业（buy-to-let）政策三十年的历程，指出租金和租户需求上升的同时，房东正承受越来越大的政策与市场压力。
As Jon Gray recalls it, Blackstone Inc.’s first foray into India started slowly and then ran into headwinds during the global financial crisis.
中文摘要 黑石集团负责人Jon Gray表示，印度正接近吸引外国资本流入的临界点。他回顾称黑石最初进入印度进展缓慢，并在全球金融危机期间遭遇逆风。
A new equity loan scheme could help people get on the property ladder, though the government might struggle to afford it
中文摘要 英国《金融时报》报道，新的股票贷款计划可能有助于购房者「跨入」房产市场，使不止房屋建造商受益，但政府或面临财政负担能力问题。
The country has almost limitless space with which to generate renewable electricity
中文摘要 英国《金融时报》指出，澳大利亚在争夺AI算力资源的竞赛中握有一张「秘密武器」——近乎无限的可用土地来建设可再生能源发电设施。
4 回复 · 程序员 节点
25 回复 · 程序员 节点
16 回复 · 程序员 节点
10 回复 · Apple 节点
15 回复 · Apple 节点
4 回复 · 程序员 节点
6 回复 · 程序员 节点
4 回复 · Apple 节点
17 回复 · Apple 节点
15 回复 · Apple 节点
该源今日无内容。