每日简报

2026-06-26

← 历史归档

google-labs-code/design.md

TypeScript · ★ 19,158 · 🍴 1,645 · 📈 1,475 stars today

A format specification for describing a visual identity to coding agents. DESIGN.md gives agents a persistent, structured understanding of a design system.

中文介绍 定义 DESIGN.md 格式,让 AI 编码代理持久、结构化地理解设计系统。适用于需要保持视觉一致性的开发团队,将设计规范直接融入代理工作流。

calesthio/OpenMontage

Python · ★ 22,039 · 🍴 2,474 · 📈 3,434 stars today

World's first open-source, agentic video production system. 12 pipelines, 52 tools, 500+ agent skills. Turn your AI coding assistant into a full video production studio.

中文介绍 全球首个开源代理化视频制作系统,内置 12 条管线、52 个工具和 500+ 代理技能。可将 AI 编码助手转变为全功能视频制作工作室,适用于自动生成视频内容。

xbtlin/ai-berkshire

Python · ★ 1,844 · 🍴 296 · 📈 309 stars today

AI 时代的伯克希尔:基于 Claude Code 的价值投资研究框架。巴菲特·芒格·段永平·李录四大师方法论 + 多Agent并行研究。| AI-era Berkshire: a value investing research framework built on Claude Code. 4 masters' methodologies + multi-agent adversarial analysis.

中文介绍 基于 Claude Code 的价值投资研究框架,融合巴菲特、芒格、段永平、李录四大师方法论,通过多 Agent 对抗研究提升分析深度。适合量化投资者和基本面研究者。

mauriceboe/TREK

TypeScript · ★ 6,635 · 🍴 606 · 📈 241 stars today

A self-hosted travel/trip planner with real-time collaboration, interactive maps, PWA support, SSO, budgets, packing lists, and more.

中文介绍 自托管的旅行规划器,支持实时协作、交互式地图、PWA、SSO、预算管理和打包清单。适合团队或个人规划行程,注重隐私和数据控制。

apple/container

Swift · ★ 43,200 · 🍴 1,268 · 📈 1,351 stars today

A tool for creating and running Linux containers using lightweight virtual machines on a Mac. It is written in Swift, and optimized for Apple silicon.

中文介绍 用 Swift 编写的工具,在 Mac 上通过轻量级虚拟机创建和运行 Linux 容器,针对 Apple Silicon 优化。为 macOS 开发者提供接近原生的容器体验。

JCodesMore/ai-website-cloner-template

TypeScript · ★ 20,409 · 🍴 3,007 · 📈 1,024 stars today

Clone any website with one command using AI coding agents

中文介绍 利用 AI 编码代理一键克隆任意网站。只需一条命令即可复制网站结构和内容,适合快速搭建原型或离线备份。

every-app/open-seo

TypeScript · ★ 2,502 · 🍴 296 · 📈 57 stars today

Open source alternative to Semrush and Ahrefs

中文介绍 Semrush 和 Ahrefs 的开源替代品,提供 SEO 分析功能。适合预算有限的中小网站运营者,自主掌控数据和工具。

garrytan/gstack

TypeScript · ★ 115,786 · 🍴 17,168 · 📈 767 stars today

Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA

中文介绍 还原 Garry Tan 的 Claude Code 配置,包含 23 个专业工具,分别扮演 CEO、设计师、工程经理、发布经理、文档工程师和质量保证角色。适合探索 AI 辅助全流程开发。

aws/agent-toolkit-for-aws

Python · ★ 1,126 · 🍴 114 · 📈 47 stars today

Official, AWS-supported MCP servers, skills, and plugins to help AI agents build on AWS

中文介绍 AWS 官方支持的工具包,提供 MCP 服务器、技能和插件,帮助 AI 代理在 AWS 上构建应用。适合使用 AWS 的开发者快速集成代理能力。

mukul975/Anthropic-Cybersecurity-Skills

Python · ★ 21,216 · 🍴 2,447 · 📈 571 stars today

817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains ·

中文介绍 817 个结构化网络安全技能,映射到 MITRE ATT&CK、NIST CSF 2.0 等 6 个框架。适用于 AI 代理安全分析、自动化攻防演练。

alibaba/page-agent

TypeScript · ★ 19,791 · 🍴 1,706 · 📈 163 stars today

JavaScript in-page GUI agent. Control web interfaces with natural language.

中文介绍 JavaScript 编写的页面内 GUI 代理,支持用自然语言操作网页。无需编写选择器,适合自动化测试、数据采集等场景。

IceWhaleTech/CasaOS

Go · ★ 34,811 · 🍴 1,992 · 📈 223 stars today

CasaOS - A simple, easy-to-use, elegant open-source Personal Cloud system.

中文介绍 简单易用的开源个人云系统,提供直观界面管理文件、应用和设备。适合家庭或小团队搭建私有云存储和服务。

opendatalab/MinerU

Python · ★ 69,553 · 🍴 5,885 · 📈 644 stars today

Transforms complex documents like PDFs and Office docs into LLM-ready markdown/JSON for your Agentic workflows.

中文介绍 将 PDF、Office 等复杂文档转换为 LLM 就绪的 Markdown 或 JSON,便于接入代理工作流。适合需要批量处理文档以训练或查询 AI 的用户。

Free-TV/IPTV

Python · ★ 18,200 · 🍴 2,678 · 📈 139 stars today

M3U Playlist for free TV channels

中文介绍 提供免费电视频道的 M3U 播放列表,覆盖多个国家。适合 IPTV 用户自行获取直播源,无需付费订阅。

shanraisshan/claude-code-best-practice

HTML · ★ 60,516 · 🍴 6,063 · 📈 287 stars today

from vibe coding to agentic engineering - practice makes claude perfect

中文介绍 从“氛围编码”到“代理式工程”的实践指南,帮助用户通过练习让 Claude Code 表现更完美。适合希望提升 AI 编码效率的开发者。

NanmiCoder/MediaCrawler

Python · ★ 52,737 · 🍴 10,920 · 📈 398 stars today

小红书笔记 | 评论爬虫、抖音视频 | 评论爬虫、快手视频 | 评论爬虫、B 站视频 | 评论爬虫、微博帖子 | 评论爬虫、百度贴吧帖子 | 百度贴吧评论回复爬虫 | 知乎问答文章|评论爬虫

中文介绍 支持小红书、抖音、快手、B 站、微博、百度贴吧、知乎等多平台笔记/视频/评论爬虫。适用于社交媒体数据采集与分析。

Are We Ready For An Agent-Native Memory System?

👍 88

Memory for large language model (LLM) agents has rapidly evolved from simple retrieval-augmented mechanisms into a data management system that supports persistent information storage, retrieval, update, consolidation, and dynamic lifecycle governance throughout agent execution. Despite this evolutio

中文介绍 探讨 LLM 智能体记忆系统从简单检索增强到支持持久存储、更新、整合与动态生命周期治理的数据管理系统的演变,分析当前系统是否达到「智能体原生」内存的要求,指出关键挑战与未来方向。

NatureBench: Can Coding Agents Match the Published SOTA of Nature-Family Papers?

👍 55

We introduce NatureBench, a cross-discipline benchmark of 90 tasks distilled from peer-reviewed Nature-family publications, designed to evaluate whether AI coding agents can move beyond reproduction toward discovery on real scientific problems. NatureBench is built on NatureGym, an automated pipelin

中文介绍 提出跨学科基准 NatureBench,包含 90 个来自 Nature 系列论文的任务,用于评估 AI 编码智能体能否超越复现、实现真正的科学发现。基于 NatureGym 自动化流水线构建。

Escaping the Self-Confirmation Trap: An Execute-Distill-Verify Paradigm for Agentic Experience Learning

👍 48

Experience-driven self-evolution is critical for large language model (LLM) agents to improve through open-world interaction. However, existing experience learning methods mostly rely on single-agent loops, where the same agent executes tasks, summarizes outcomes, and determines memory content. This

中文介绍 针对 LLM 智能体经验学习中的自我确认陷阱,提出「执行-蒸馏-验证」范式。将任务执行、经验总结与记忆判定分离,避免单智能体闭环导致的偏差,提升开放世界交互中的自适应能力。

Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models

👍 40

We present Wan-Streamer, a native-streaming, end-to-end interactive foundation model designed from the ground up for real-time, low-latency, full-duplex audio-visual interaction. Wan-Streamer seamlessly models language, audio, and video as both input and output within a single Transformer, where the

中文介绍 提出 Wan-Streamer,原生流式端到端交互基础模型,基于单一 Transformer 同时建模语言、音频和视频的输入与输出,实现低延迟、全双工音视频实时交互。

ShutterMuse: Capture-Time Photography Guidance with MLLMs

👍 37

Real-world photography requires capture-time guidance for both camera framing and subject pose. Yet existing aesthetic cropping benchmarks mainly evaluate post-hoc crop prediction and overlook subject-side recommendations, leaving the capture-time guidance capabilities of multimodal large language m

中文介绍 提出 ShutterMuse,利用多模态大语言模型(MLLM)在拍照时提供取景构图与主体姿态指引。不同于事后裁剪评估,该方法在拍摄阶段给出实时建议,填补了 MLLM 在拍照即时指导方面的空白。

Beyond NL2Code: A Structured Survey of Multimodal Code Intelligence

👍 27

While Large Language Models (LLMs) have substantially advanced text-to-code synthesis, many real programming tasks specify intent through visual artifacts such as screenshots, charts, vector drawings, videos, and interactive states. These tasks require models to connect visual perception to executab

中文介绍 综述多模态代码智能,超越传统 NL2Code。指出真实编程任务常通过截图、图表、视频等视觉形式表达意图,要求模型连接视觉感知与可执行代码。梳理现有方法、数据集与挑战。

AOHP: An Open-Source OS-Level Agent Harness for Personalized, Efficient and Secure Interaction

👍 27

AI agents are driving a new software paradigm, with the ability to autonomously call tools, extract information, manage memory, and complete tasks that span applications and data sources. Most existing end-user operating systems, however, are designed for application-centric workflows and offer litt

中文介绍 开源操作系统级智能体框架 AOHP,支持个性化、高效、安全的交互。提供工具调用、信息提取、内存管理等功能,打破应用壁垒,实现跨应用与数据源的自主任务完成。

MVTrack4Gen: Multi-View Point Tracking as Geometric Supervision for 4D Video Generation

👍 26

Synthesizing a novel-view video from a monocular reference video along a target camera trajectory requires both geometric consistency and motion fidelity with respect to the reference video. Existing methods based on explicit 3D representations are limited by the accuracy of off-the-shelf reconstruc

中文介绍 提出 MVTrack4Gen,将多视角点跟踪作为几何监督信号用于 4D 视频生成。从单目参考视频合成新视角视频,利用点轨迹约束保证几何一致性与运动保真度,克服显式 3D 表示依赖准确重建的局限。

LingxiDiagBench: A Multi-Agent Framework for Benchmarking LLMs in Chinese Psychiatric Consultation and Diagnosis

👍 21

Mental disorders are highly prevalent worldwide, but the shortage of psychiatrists and the inherent subjectivity of interview-based diagnosis create substantial barriers to timely and consistent mental-health assessment. Progress in AI-assisted psychiatric diagnosis is constrained by the absence of

中文介绍 提出多智能体框架 LingxiDiagBench,用于评估 LLM 在中文精神科咨询与诊断中的表现。模拟问诊流程,缓解精神科医生短缺与主观性障碍,为 AI 辅助精神健康评估提供标准化基准。

UnityShots: Memory-Driven Multi-Shot Audio-Video Generation with Boundary-Aware Gating

👍 19

Generating a coherent multi-shot video requires structured cross-shot memory. Subject appearance, scene context, and speaker identity must persist across cuts. Existing approaches either train end-to-end over fixed-length sequences and cannot scale, generate shot-by-shot with memory banks that grow

中文介绍 提出 UnityShots,基于结构化跨镜头记忆的多片段音视频生成方法。采用边界感知门控机制,在镜头切换时保持主题外观、场景上下文和说话者身份的一致性,解决现有方法无法缩放或记忆膨胀问题。

V-Zero: Answer-Label-Free On-Policy Distillation with Contrastive Evidence Gating for Fine-Grained Visual Reasoning

👍 19

Fine-grained visual reasoning requires multimodal large language models (MLLMs) to identify task-relevant visual evidence and ground their reasoning in local image regions. Existing agentic methods typically rely on reinforcement learning with verifiable rewards or supervised fine-tuning on large-sc

中文介绍 提出 V-Zero,无需答案标签的在策略蒸馏方法,结合对比证据门控,提升 MLLM 在细粒度视觉推理中的局部区域定位与证据链能力。避免依赖 RL 或大规模监督微调。

FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generation

👍 18

Generating explorable 3D scenes from a single image requires strong generative priors and accurate geometric representations suitable for downstream use. Current video diffusion models offer high-quality generation and implicitly encode multi-view geometric structure in latent space. However, existi

中文介绍 提出 FLAT(前馈潜在三角形喷溅),从单张图像生成几何精确的 3D 场景。利用视频扩散模型中隐含的多视图几何先验,通过三角形喷溅表示实现高效、可探索的 3D 重建。

Critique of Agent Model

👍 16

What is an agent? What constitutes agency? With the rise of Large Language Model (LLM) systems marketed as ``coding agents'', ``AI co-scientists'', and other ``agentic" tools that promise to drive up productivity, and at the same time, ``existential" concerns such as AI escaping human control with d

中文介绍 对 Agent 模型进行批判性分析,探讨智能体的本质与构成。质疑当前 LLM 系统(如「编码智能体」「AI 合作科学家」)的「智能体」宣称,厘清自主性与工具性界限。

Semantic Browsing: Controllable Diversity for Image Generation

👍 15

Modern text-to-image models excel in visual fidelity and prompt adherence. However, this strict adherence comes at the cost of diversity: generated samples tend to collapse into a single visual interpretation. Existing methods to improve diversity produce outputs driven by incidental variations rath

中文介绍 提出 Semantic Browsing,在文本到图像生成中实现可控多样性。允许用户通过语义空间浏览不同视觉解释,而非依赖随机变异,在保持提示一致性的前提下增加生成结果的丰富性。

IV-CoT: Implicit Visual Chain-of-Thought for Structure-Aware Text-to-Image Generation

👍 14

Unified multi-modal large language models (MLLMs) have achieved strong text-to-image generation quality, but still struggle with structure-aware prompt following, where object counts, spatial relations, attribute bindings, and coarse layouts must be preserved. We attribute this limitation in part to

中文介绍 提出隐式视觉思维链 IV-CoT,增强 MLLM 文本到图像生成中的结构感知遵循能力。通过隐式视觉推理步骤处理对象计数、空间关系、属性绑定和粗略布局,提升生成图像与复杂提示的匹配度。

The Hitchhiker's Guide to Agentic AI: From Foundations to Systems

👍 11

The Hitchhiker's Guide to Agentic AI is a comprehensive practitioner's reference for building autonomous AI systems. The book covers the full stack from first principles to production deployment, organized around a central thesis: building great agentic systems requires understanding every layer of

中文介绍 面向实践者的智能体 AI 综合参考书,覆盖从第一性原理到生产部署的完整技术栈。强调构建优秀智能体系统需要理解每一层抽象,包括模型、记忆、工具使用与编排。

Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot Do

👍 8

Chain-of-Thought (CoT) has become a standard method for improving reasoning capabilities in large language models (LLMs) by eliciting step-by-step thinking, but its effectiveness in multimodal tasks remains unclear. In this paper, we aim to systematically investigate the key question: What can multi

中文介绍 系统探究多模态思维链(CoT)推理的能力边界。通过实验分析 CoT 在不同多模态任务中的效果与局限,指出其提升可解释性但可能引入错误传播,并对比单模态与多模态场景下的差异。

Autodata: An agentic data scientist to create high quality synthetic data

👍 8

We introduce Autodata, a general method that enables AI agents to act as data scientists who build high quality training and evaluation data. We show how to train (meta-optimize) such a data scientist agent, so that it learns to create even stronger data. We describe the overall formulation, and a s

中文介绍 提出 Autodata,使 AI 智能体扮演数据科学家角色,自动构建高质量合成训练与评估数据。通过元优化训练数据科学家智能体,使其学会生成更强的数据,形成闭环提升。

TryOnCrafter: Unleashing Camera Trajectories for Realistic Video Virtual Try-on via a Renderable 4D Try-on Proxy

👍 6

While Video Virtual Try-on (VVT) has achieved remarkable progress in synthesizing realistic garment overlays on dynamic subjects, existing paradigms remains fundamentally constrained by a passive dependency on source camera trajectories, failing to accommodate the requisite interactive freedom for o

中文介绍 提出 TryOnCrafter,通过可渲染的 4D 试穿代理,释放相机轨迹自由度,实现真实感视频虚拟试穿。克服现有方法对源相机轨迹的被动依赖,支持交互式自由视角试穿体验。

Advancing WordArt-Oriented Scene Text Recognition: Datasets and Methods

👍 6

WordArt (artistic text) features highly customized fonts, textures, and layouts, making WordArt-oriented scene TExt Recognition (WATER) substantially more challenging than general Scene Text Recognition (STR). Existing STR datasets and methods, typically built around regular scene text and fixed-tem

中文介绍 针对艺术字场景文本识别(WATER)任务,提出新数据集与方法。由于艺术字包含高度定制化的字体、纹理和布局,现有 STR 方法与数据集表现不佳,新工作填补了该领域空白。

The efficiency gap: How do cells and GPUs compare when running the exact same algorithm?

@spandan_madan · 1.1K 粉丝 · 626.9K 阅 · 516 赞 · 28 转

AI hardware is having a moment. Hyperscaler capex on AI data centres is on track to clear $690 billion in 2026, and private equity has followed in scale — Blackstone alone reports a $55B+ data-centre

中文介绍 对比细胞与GPU执行相同算法的效率差距,背景是AI数据中心投资预计2026年达6900亿美元,Blackstone等私募跟进,指出生物计算在能效上的潜在优势。

Introducing Engram: Scaling compute on your context

@EngramLab · 1.2K 粉丝 · 255.7K 阅 · 537 赞 · 76 转

We’re Engram. We’re building AI that learns from you and deeply understands your work. Today’s AI models don’t understand what you do. Not really. Everything models know comes from their training –

中文介绍 EngramLab发布AI助手Engram,强调其能真正学习用户并理解工作上下文,而非仅依赖训练数据,实现上下文计算扩展。

30 Core Agentic Engineering Concepts Every Developer Should Know

@sairahul1 · 120.1K 粉丝 · 249.0K 阅 · 502 赞 · 90 转

There are 8 billion people on the planet. Only a fraction of developers understand how AI agents actually work. Not the demos. Not the hype. The real engineering underneath. Every week a new agent

中文介绍 sairahul1总结30个开发者必须了解的Agentic工程核心概念,涵盖底层原理而非浅层演示,帮助理解AI Agent的真实工作机制。

Why we're bullish on loops

@posthog · 21.8K 粉丝 · 162.4K 阅 · 512 赞 · 36 转

When the creators of both OpenClaw and Claude Code speak, people listen. And recently Peter Steinberger and Boris Cherny have both been talking about the same concept: loops. Their argument? You

中文介绍 PostHog看好“loops”模式,引用OpenClaw和Claude Code创始人的观点,认为循环迭代式AI工作流比单次调用更有效。

Subnets mine TAO; TAO mines Subnets

@const_reborn · 29.7K 粉丝 · 79.8K 阅 · 503 赞 · 116 转

e_i \;\propto\; \underbrace{\rho_i \times \bar{p}_i}_{\text{linear (maximize)}} \times \underbrace{(1 - b_i)}_{\text{boolean gate}} Disclaimer: this upgrade only effects subnet owners and dynamic TAO

中文介绍 const_reborn分析Bittensor网络中子网挖矿与TAO代币的互惠关系,提出升级仅影响子网所有者与动态TAO,涉及线性最大化与布尔门控机制。

Introducing computer use in Gemini 3.5 Flash

@GoogleAIStudio · 179.4K 粉丝 · 41.0K 阅 · 605 赞 · 57 转

Computer use is now a built-in tool supported in Gemini 3.5 Flash, delivering our best performance yet for agentic computer use tasks. Previously only available as a standalone Gemini 2.5 computer use

中文介绍 Google AI Studio在Gemini 3.5 Flash中内置计算机使用(computer use)工具,支持智能体操作电脑任务,性能优于此前独立版的Gemini 2.5。

Building a skill optimization loop

@zachlloydtweets · 11.4K 粉丝 · 39.2K 阅 · 504 赞 · 68 转

This post shows how to create a loop with automated feedback that an agent can run to optimize its own Skills. It uses an automated grader with computer use to assess how well a Skill is performing,

中文介绍 zachlloydtweets展示如何构建自动化反馈循环,让AI Agent通过计算机使用来自动评估并优化自身技能,实现自我提升。

How does vLLM work?

@amitiitbhu · 23.6K 粉丝 · 35.7K 阅 · 514 赞 · 63 转

In this blog, we will learn about how vLLM works. We will also see why we need it, how it manages memory so cleverly, and where it is used in the real world to serve large language models to many

中文介绍 amitiitbhu解释vLLM的工作机制,包括其为何必要、巧妙的显存管理方式,以及在现实世界中服务大语言模型的应用场景。

Run a vLLM Server on HF Jobs in One Command

中文介绍 Hugging Face博客介绍,用户可通过一条命令在HF Jobs上部署vLLM推理服务器,简化大语言模型服务部署流程。

Repositioning retail for the AI era

Artificial intelligence is rapidly reshaping retail, but not in the ways consumers might immediately notice. The biggest transformation may not be flashy virtual try-ons or chatbot shopping assistants, but in how decisions are made behind the scenes: how products surface in search results, how inven

中文介绍 MIT Tech Review指出,AI重塑零售业的主要变化并非虚拟试穿或聊天机器人,而是幕后决策方式,如产品如何出现在搜索结果中。

[AINews] It's Meta-Harness Summer

Move over, Harness Engineering, it is time for the harness of harnesses!

中文介绍 Latent Space发布AI新闻,称「Meta-Harness夏季」到来,强调超越传统Harness Engineering,进入更高级的「harness of harnesses」阶段。

How agents are transforming work

A new OpenAI research paper shows how AI agents are transforming work, enabling longer, more complex tasks and expanding productivity across roles.

中文介绍 OpenAI发布研究论文,展示AI代理如何改变工作方式,支持更长时间、更复杂的任务,并提升各角色的生产力。

The emergence of the web data infrastructure layer for AI

AI is booming. New use cases are emerging each day. To capitalize on the technology’s potential, enterprises require data at scale. In many cases, though, the relevant information is blocked or unstructured, which limits its use by AI models. To understand this challenge, consider the foundation of

中文介绍 MIT Tech Review探讨AI时代网络数据基础设施层的崛起,企业需大规模数据,但信息常被封锁或非结构化,限制AI模型使用。

OpenAI and Broadcom unveil LLM-optimized inference chip

OpenAI and Broadcom introduce Jalapeño, a custom AI chip built for LLM inference to improve performance, efficiency, and scale across AI systems.

中文介绍 OpenAI与Broadcom推出定制AI芯片Jalapeño,专为大语言模型推理优化,旨在提升性能、效率和扩展性。

not much happened today

**OpenAI** announced **Jalapeño**, its first custom AI chip for LLM inference, built with **Broadcom**, aiming to control more of the AI stack and improve compute economics with a fast 9-month design cycle. Community analysis suggests Jalapeño features **216GB HBM3E**, **~7.1–7.4 TB/s bandwidth**, a

中文介绍 Smol AI News报道:OpenAI与Broadcom发布定制推理芯片Jalapeño,设计周期仅9个月;社区分析指出其配备216GB HBM3E内存,算力约7.1-7.4 TFLOPS。

市场总览

当前市场整体技术面偏弱,风险资产承压明显。美股方面,SPY、QQQ虽维持多头排列但价格均跌破SMA20,RSI回落至46-50区间,短期动能衰减;科技龙头如MSFT、META、TSLA等已形成空头排列,RSI接近或进入超卖区。加密市场极度恐慌,恐慌贪婪指数仅13,BTC、ETH、SOL均处于空头排列且RSI在30-40之间,价格远低于中长期均线。中概股跌势惨烈,BABA、PDD、JD的RSI分别跌至16.7、24、24.4,超卖信号强烈,空头排列明确。商品外汇方面,黄金、原油同步下行,GC=F、CL=F的RSI分别为32.5和28.7,MACD死叉;美元指数DXY则一枝独秀,RSI达74.5超买,多头排列且接近52周高位,对新兴市场资产形成压制。

今日关注

DX-Y.NYB 美元指数 DXY
偏上行

RSI 14 达 74.5 处于超买区,价格高于所有均线(SMA20=100.03, SMA50=99.16, SMA200=98.75),形成多头排列,且距 52 周高点仅 -0.29%。MACD 柱线正值(0.6047)且信号线(0.4318)上行,显示强劲上行趋势。

MSFT 微软 (MSFT)
偏下行

RSI 14 为 28.7 进入超卖区,价格远低于所有均线(SMA20=402.81, SMA50=411.74, SMA200=448.61),形成空头排列。MACD 柱线 -14.23 远低于信号线 -8.80,死叉持续扩散,距 52 周低点仅 1.04%,下行压力显著。

BABA 阿里巴巴 (BABA)
偏下行

RSI 14 为 16.7 深度超卖,价格跌破所有均线(SMA20=115.1, SMA50=127.06, SMA200=148.48)且空头排列明确。MACD 柱线 -8.10 较信号线 -6.37 加速下行,距 52 周低点仅 0.37%,技术面极度弱势。

SPY S&P 500 ETF
中性

RSI 14 为 46.5 处于中性区域,价格介于 SMA20(744.88)与 SMA50(733.77)之间,但仍高于 SMA200(690.14),整体多头排列格局未破。MACD 柱线 0.83 虽为正但低于信号线 3.36,短期动能偏弱,多空胶着。

全部资产

^VIX

VIX 恐慌指数

$18.89 +1.40%
5 日
+2.44%
距 52w 高
-46.5%
RSI(14)
52.6
趋势
中性
SMA 20 / 50 / 200
17.79 / 17.78 / 18.63
MACD / 信号
0.143 / 0.009
MACD 金叉 (2 天前)

^TNX

10Y 美债收益率 (%)

$4.39 -1.33%
5 日
-2.12%
距 52w 高
-12.1%
RSI(14)
41.5
趋势
中性
SMA 20 / 50 / 200
4.48 / 4.43 / 4.22
MACD / 信号
-0.003 / 0.013

DX-Y.NYB

美元指数 DXY

$101.51 -0.10%
5 日
+1.42%
距 52w 高
-0.3%
RSI(14)
74.5
趋势
多头
SMA 20 / 50 / 200
100.03 / 99.16 / 98.75
MACD / 信号
0.605 / 0.432
RSI 超买接近 52 周高多头排列

SPY

S&P 500 ETF

$734.30 +0.14%
5 日
-0.90%
距 52w 高
-3.4%
RSI(14)
46.5
趋势
多头
SMA 20 / 50 / 200
744.88 / 733.77 / 690.14
MACD / 信号
0.829 / 3.356
多头排列

QQQ

Nasdaq 100 ETF

$716.38 +0.81%
5 日
-0.85%
距 52w 高
-4.3%
RSI(14)
49.9
趋势
多头
SMA 20 / 50 / 200
726.22 / 701.41 / 631.52
MACD / 信号
5.105 / 8.563
多头排列

AAPL

Apple

$275.15 -6.12%
5 日
-7.03%
距 52w 高
-13.3%
RSI(14)
32.2
趋势
中性
SMA 20 / 50 / 200
299.73 / 291.15 / 269.21
MACD / 信号
-1.588 / 1.199

MSFT

Microsoft

$352.83 -3.46%
5 日
-6.88%
距 52w 高
-36.5%
RSI(14)
28.7
趋势
空头
SMA 20 / 50 / 200
402.81 / 411.74 / 448.61
MACD / 信号
-14.234 / -8.797
RSI 超卖接近 52 周低空头排列

NVDA

Nvidia

$195.74 -1.64%
5 日
-4.35%
距 52w 高
-17.2%
RSI(14)
39.6
趋势
中性
SMA 20 / 50 / 200
208.85 / 210.21 / 190.53
MACD / 信号
-2.940 / -1.317

GOOGL

Alphabet

$343.71 -0.46%
5 日
-5.52%
距 52w 高
-15.9%
RSI(14)
36.2
趋势
中性
SMA 20 / 50 / 200
363.45 / 369.27 / 313.36
MACD / 信号
-6.254 / -3.301

TSLA

Tesla

$375.12 -0.11%
5 日
-5.36%
距 52w 高
-24.8%
RSI(14)
38.9
趋势
空头
SMA 20 / 50 / 200
404.67 / 405.01 / 417.80
MACD / 信号
-7.383 / -3.473
空头排列

META

Meta

$542.87 -2.65%
5 日
-4.35%
距 52w 高
-31.8%
RSI(14)
34.1
趋势
空头
SMA 20 / 50 / 200
587.54 / 615.32 / 651.10
MACD / 信号
-15.724 / -12.405
空头排列
加密恐慌贪婪
13
极度恐慌
加密总市值
$2.15 T
-1.29% / 24h
BTC 主导率
55.8%
ETH 8.8%
24h 成交量
$96.3 B
活跃币 17,387

BTC-USD

Bitcoin

$59,752.38 -2.04%
5 日
-6.99%
距 52w 高
-52.7%
RSI(14)
30.5
趋势
空头
SMA 20 / 50 / 200
63,255.87 / 70,631.58 / 76,154.51
MACD / 信号
-2,217.401 / -2,288.787
接近 52 周低空头排列

ETH-USD

Ethereum

$1,567.68 -3.23%
5 日
-9.87%
距 52w 高
-68.4%
RSI(14)
30.7
趋势
空头
SMA 20 / 50 / 200
1,686.13 / 1,934.92 / 2,335.38
MACD / 信号
-74.438 / -78.022
空头排列

SOL-USD

Solana

$67.94 -0.06%
5 日
-7.15%
距 52w 高
-73.2%
RSI(14)
40.1
趋势
空头
SMA 20 / 50 / 200
68.94 / 78.42 / 96.11
MACD / 信号
-1.906 / -2.326
空头排列

BABA

阿里巴巴 (BABA)

$95.07 -4.74%
5 日
-11.51%
距 52w 高
-50.7%
RSI(14)
16.7
趋势
空头
SMA 20 / 50 / 200
115.10 / 127.06 / 148.48
MACD / 信号
-8.099 / -6.373
RSI 超卖接近 52 周低空头排列

PDD

拼多多 (PDD)

$73.30 -3.22%
5 日
-8.21%
距 52w 高
-47.4%
RSI(14)
24.0
趋势
空头
SMA 20 / 50 / 200
81.84 / 91.76 / 109.62
MACD / 信号
-4.636 / -4.210
RSI 超卖接近 52 周低空头排列

JD

京东 (JD)

$25.19 -1.14%
5 日
-9.75%
距 52w 高
-31.7%
RSI(14)
24.4
趋势
空头
SMA 20 / 50 / 200
28.17 / 29.79 / 30.14
MACD / 信号
-1.103 / -0.770
RSI 超卖接近 52 周低空头排列

0700.HK

腾讯控股 (0700.HK)

HK$421.40 -1.73%
5 日
-5.39%
距 52w 高
-38.3%
RSI(14)
38.7
趋势
空头
SMA 20 / 50 / 200
446.25 / 463.08 / 562.43
MACD / 信号
-9.529 / -6.955
MACD 死叉 (3 天前)接近 52 周低空头排列

GC=F

黄金期货

$4,029.70 +0.99%
5 日
-7.55%
距 52w 高
-27.9%
RSI(14)
32.5
趋势
中性
SMA 20 / 50 / 200
4,292.84 / 4,502.51 / 4,446.03
MACD / 信号
-121.638 / -102.896
MACD 死叉 (3 天前)

CL=F

WTI 原油期货

$71.57 +1.75%
5 日
-6.80%
距 52w 高
-40.1%
RSI(14)
28.7
趋势
中性
SMA 20 / 50 / 200
84.20 / 92.42 / 73.86
MACD / 信号
-6.298 / -4.873
RSI 超卖

USDCNY=X

美元 / 人民币

¥6.79 -0.01%
5 日
+0.47%
距 52w 高
-5.9%
RSI(14)
56.7
趋势
空头
SMA 20 / 50 / 200
6.77 / 6.79 / 6.95
MACD / 信号
-0.005 / -0.009
接近 52 周低空头排列
风险提示

过去走势不代表未来表现,以上技术指标解读仅供参考,不构成任何投资建议。市场存在不确定性,请结合自身风险承受能力独立决策。

Officials Shut Down Kenya’s Capital to Block Protest

The protest, an annual event, has become a showcase of the growing political power of Kenya’s youth.

中文摘要 肯尼亚官员关闭首都内罗毕以阻止抗议活动。该抗议为年度事件,凸显了该国青年日益增强的政治力量。

Australia news live: PM unsure if Australians caught up in Venezuela quake; new warning over fire ant threat to pets

Prime minister says it is ‘possible’ that Australians are affected by severe earthquake. Follow today’s news live LIVE: Australia vs Paraguay – Socceroos World Cup game Get our breaking news email, free app or daily news podcast Albanese says Stefanovic saga ‘unfortunate’, but ‘words do matter’ The

中文摘要 澳大利亚总理表示可能有澳人在委内瑞拉地震中受影响。另就火蚁对宠物威胁发布新警告。

Paris restricts alcohol consumption and sales as Europe's heatwave shifts east

Temperatures in Germany could hit 40C in some areas while French officials bring in alcohol restrictions to ease pressure on hospitals overwhelmed by patients during heatwave.

中文摘要 欧洲热浪东移,巴黎限制酒精消费和销售以缓解医院压力。德国部分地区气温可能达40摄氏度。

Venezuela Live Updates: Search for Survivors Grows More Urgent 24 Hours After Quakes

Back-to-back earthquakes left at least 188 dead and injured more than 1,500 people, the authorities said. Residents of the port city of La Guaira estimated that hundreds of people were trapped in rubble. The toll was expected to rise.

中文摘要 委内瑞拉连续地震已造成至少188人死亡、超1500人受伤。港口城市拉瓜伊拉数百人被困废墟,预计死亡人数将继续上升。

Experts say three recent powerful earthquakes are not related

5.6-magnitude California tremor, 7.2-magnitude temblor in Japan and two quakes in Venezuela all struck within hours A 5.6-magnitude earthquake struck a rural part of northern California on Wednesday. Hours later, a 7.2-magnitude earthquake hit the northern coast of Japan and two powerful earthquakes

中文摘要 专家称近期三次强震不相关:加州5.6级地震、日本7.2级地震及委内瑞拉两次地震均在数小时内发生,但彼此无关。

Who is Tommy Robinson? The Karl Stefanovic guest who has cost Australia’s famed TV host his job

Robinson, whose real name is Stephen Yaxley-Lennon, is an anti-Islamic, far-right political activist well known in the UK and Europe Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast The man who calls himself Tommy Robinson has roared

中文摘要 汤米·罗宾逊(本名斯蒂芬·亚克斯利-列侬)是英国及欧洲知名的反伊斯兰极右翼政治活动家。他作为嘉宾参加澳大利亚主持人卡尔·斯特凡诺维奇的节目,导致后者失业。

Why the trail went cold in Nancy Guthrie case

The lack of a breakthrough in the abduction of a famous TV anchor's mother is puzzling given the huge publicity.

中文摘要 南希·格思里(著名电视主播的母亲)绑架案缺乏突破,尽管获得大量媒体报道,仍无进展,令人费解。

‘No offence detected’: Victoria police close investigation into Moira Deeming allegation of assault by Matthew Guy

Guy accuses Deeming of making ‘ridiculous and unsubstantiated’ claims as state Liberal party grapples with fallout Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast Police have concluded their investigation into Victorian Liberal MP Mo

中文摘要 维多利亚州警方结束对莫伊拉·迪明指控马修·盖伊袭击的调查,称“未发现犯罪行为”。盖伊指责迪明提出“荒谬且无根据”的指控。

Congo Ebola Crisis: Contact Tracing Is Dangerously Behind, Officials Warn

Most of the people testing positive for Ebola in the Democratic Republic of Congo are not on health workers’ radar, suggesting that contact tracing is lagging dangerously behind.

中文摘要 官员警告刚果(金)埃博拉危机中接触者追踪严重滞后。大多数检测阳性者未被卫生工作者掌握。

How Long Can a Person Survive Under Rubble?

It depends on water, air and how quickly rescuers are able to find people.

中文摘要 人在废墟下能存活多久取决于水源、空气及救援速度。

Iran Strikes Ship in Strait of Hormuz, Undermining Efforts to Restore Traffic

The attack, confirmed by U.S. and Iranian officials, came after Tehran threatened to disrupt shipping in the waterway and as Washington sought regional support for its peace agreement.

中文摘要 伊朗在霍尔木兹海峡袭击船只,破坏恢复通航努力。美伊官员证实袭击。此前德黑兰威胁扰乱航运,华盛顿寻求地区支持其和平协议。

UN agency pauses ship evacuations through strait of Hormuz after vessel struck

International Maritime Organization says safety guarantees must be confirmed before ships can move again A United Nations agency has paused the evacuation of ships through the strait of Hormuz after the British military said a vessel was hit by a projectile off the coast of Oman following the passag

中文摘要 联合国机构暂停霍尔木兹海峡船舶疏散。英国军方称一艘船只在阿曼沿海遭炮弹击中,国际海事组织表示需先确认安全保证。

Oil Prices Rise as Ship Struck in Hormuz

Oil prices turned higher after a ship was hit by an unknown projectile in the Strait of Hormuz, sustaining damage to the bridge. Bloomberg's Mica Soellner has the context. (Source: Bloomberg)

中文摘要 一艘船只在霍尔木兹海峡遭不明射弹击中,桥梁受损,油价随之走高。

Gold Steadies Near $4,000 as Inflation Data Eases Rate-Hike Bets

Gold steadied near $4,000 an ounce as the latest US inflation gauge tempered expectations for an interest-rate hike, capping a tumultuous week that saw bullion slump to the lowest since November.

中文摘要 金价稳定在每盎司4000美元附近,因最新美国通胀指标缓和了加息预期,此前一周金价曾跌至11月以来最低。

Warning over power bank fire risk on flights as summer holidays begin

Lithium battery fires are the number one safety risk to aircraft, yet the number of devices found in hold bags has nearly doubled in a year.

中文摘要 锂电池火灾是飞机上首要安全风险,一年来托运行李中被发现的此类设备数量几乎翻倍,暑假出行需警惕。

BHP’s Incoming CEO Craig Tweaks Leadership Team Before Start

BHP Group’s incoming Chief Executive Officer Brandon Craig has announced changes to the company’s leadership team before he officially assumes the role on July 1.

中文摘要 必和必拓候任CEO布兰登·克雷格在7月1日正式上任前宣布调整公司领导团队。

Teens who hacked TfL were known to police years before cyber-attack

Owen Flowers and Thalha Jubair were convicted for their roles in the attack, which led to large costs for Transport for London.

中文摘要 两名青少年因入侵伦敦交通局系统被判刑,该网络攻击造成巨额损失,而警方多年前就已认识他们。

Apple Supplier Lingyi Set for HK Debut After $1.1 Billion Float

Lingyi iTech Guangdong Co. a Chinese maker of electronic devices, is set to begin trading in Hong Kong on Friday after raising HK$8.3 billion ($1.06 billion) in a share sale, part of the city’s busiest month for listings this year.

中文摘要 苹果供应商广东领益智造在港募资83亿港元(约10.6亿美元)后,将于周五在港交所挂牌,这是香港今年以来最繁忙的上市月份之一。

Attack in Strait of Hormuz halts evacuation plan for stranded ships

International Maritime Organization announces pause after container ship was hit by ‘unknown projectile’

中文摘要 国际海事组织宣布暂停疏散计划,此前一艘集装箱船在霍尔木兹海峡遭“不明射弹”击中。

The Close 6/25/2026

Bloomberg Television brings you the latest news and analysis leading up to the final minutes and seconds before and after the closing bell on Wall Street. Today's guests are Citi Wealth Head of Portfolio Construction & Analytics Olaolu Aganga, Moderna CEO Stéphane Bancel, 248 Ventures Chief Investme

中文摘要 该节目为华尔街收盘报道,嘉宾包括花旗财富投资组合构建主管Olaolu Aganga、Moderna CEO Stéphane Bancel等。

Wall Street Embraces the Dollar as Warsh’s Fed Activates Bulls

The dollar is wrapping up one of its best months in a year as a raft of Wall Street banks see a turnaround of fortunes for the US currency.

中文摘要 美元即将迎来年来最佳月度表现之一,多家华尔街银行看好美元前景。

气笑了,兄弟们,第一次见API密钥管理面板变成评论区的,所谓的低价opencode其实本质上就是一号多卖,别买了兄弟们,大半夜的,还是绷不住笑了

刚打算给新的任务弄个GLM5.2的API给新的任务对接上,然后美美睡觉,结果发现好热闹,我买的是给邮箱的,但我说实话,就算是你给支付链接也差不多,很快就会被封的,太变态了,API评论区越来越多受害者了 这个店铺已经被迫关停了 133 个帖子 - 76 位参与者 阅读完整话题

【哲のGrok&Gemini】魂兮归来,无限Grok、Gemini重新复活,服务器升级超级大鸡8c32g,kimi、glm预告

从 【哲のgrok】今天的grok和无限Gemini、grok key的使用反馈贴 - 福利羊毛 / 福利羊毛, Lv1 - LINUX DO and 【哲のgrok】今日一百个普号+限时无限量grok、Gemini(方便大家直接用 - 福利羊毛 / 福利羊毛, Lv1 - LINUX DO and 【哲のgrok&Gemini】之前的无限grok、Gemini大家蹬的如何?顺便奉上这次的限时无限量 - 福利羊毛 - LINUX DO and 【哲のgrok&Gemini】今日一百个普号+无限量Grok、Gemini自取【已全复活】 - 福利羊毛 / 福利羊毛, Lv1 - LINUX DO

GGgrok公益站正式开业!(为了其他佬友的体验,展示关闭注册)

本帖使用社区公益推广,符合推广要求。我申明并遵循社区要求的以下内容: 我的项目是免费使用的,无收费(变相收费、赞助)部分: 是 我的帖子已经打上 公益推广 标签: 是 我的项目属于个人项目,与公司或商业机构无关: 是 我的项目不存在QQ、TG等群组引流: 是 我的项目不存在非运营必要的网站引流: 是 / 否 我的项目不存在为他人推广、AFF: 是 我的项目无关联的商业项目: 是 我的站点存在登录,并已接入 LINUX DO Connect: 是 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是 以上选择我承诺是永久有效的,接受社区和佬友监督: 是 以下为项目介绍正文内容,AI生成、润

『富可敌国』『高级推广』 上架 海外版本 dreamina-seedance-2-0-mini

从 『富可敌国』『高级推广』只需要一个创意,剩下的交给星辰~即梦,创造你的梦~ 继续讨论: 海外版本即梦全量上线 dreamina-seedance-2-0-260128 dreamina-seedance-2-0-fast-260128 dreamina-seedance-2-0-mini-260615 价格 一秒大约1人民币,具体计费完全基于 total_token计算,和火山一致 我们的优势 不用排队 审核围栏低 并发高 支持对公转账和企业对接 进入 Seedance · AI 视频生成 使用文档 ShowDoc 官网 New API 技术支持 30 个帖子 - 25 位参与者 阅读完整

同事炒股入迷耽误工作,该咋办?

事情是这样的,我们是未上市的国资企业,同事小我几岁,当时是应届生进的公司,现在已经来了4年了,他们同期的小伙伴都走的差不多了。 他从24年开始上班炒股,然后加班炒股卷绩效,公司绩效跟工时正相关,所以他绩效一直还不错。前面还好,工作也能按时完成,也能出差去客户现场抗压力。 到今年开始愈演愈烈,上班就一直在看手机,安排的工作也是很敷衍的完成。上班时间喊他也是爱理不理,休息时间跟他说话他就跟我说他今天赚了几千,几万,说今年已经赚了一辆su7了。从他口里感觉这些工作都没意义,AI代替的也很快,35岁被裁一场空,卷了半天不如他一天的收入。现在就等着公司裁员拿赔偿专心炒股。现在班也不加了,出问题也找不到人

随手写--分享我的量化之路

先说我目前的情况,目前手里有两个正在实盘的策略,低频的收益过去30天收益14%,中频的收益过去30天20%。 单纯是个人散修玩量化(币圈),研学了差不多两年,从合约到杠杆到爆仓,最后选择避免情绪交易,一切交给程序。所以如果有佬,希望嘴下留情。 学习路线1:b站学量化基础概念–>youtube学看别人量化策略–>先复现别人的策略(我复现了7个)–>在Tradingview上看回测结果–>利用AI优化策略–>放弃所有策略。 当自己尝试了很多别人的策略以后会发现,真正赚钱的策略人家肯定不会公开,他们公开的必定只是部分,可能会存在一些指标参数or隐藏了关键逻辑,所以还是得研发自己的策略。 学习路线2:

Pixel手机使用不负责推荐

大家都在聊怎么薅Pixel号,但是正儿八经(并不是)聊Pixel手机的帖子好像没有。作为从Nexus一路过来的活化石,我感觉这个重任落在了我肩上(自动滑肩:) 来到6月的尾巴,谷歌好像泄了气,Gemini 3.5 Pro难产,Antigravity开始摆烂。 去年的美国甜心,今年被调侃是美国豆包。这个AI时代,强者为王,也抢着为王,慢一步就是海量的用户流失。 忠诚度不如上下文长度。 但是谷歌毕竟还是互联网老钱,搜索,邮箱,YouTube,相册,大把你用了就离不开的产品。 比如我就离不开Pixel手机,因为上面所有服务都和Pixel完美契合。还有更好用的Gemini(端侧模型),免费的你懂的,纯

女朋友说我脑袋里整天不是屎就是黄

在站内看见有意思的NSFW帖就展示给女朋友看来着 结果她说我脑袋里整天不是屎就是黄 转念一想还真是 我脑子里全是始皇 67 个帖子 - 62 位参与者 阅读完整话题

把B站,小红书、抖音推荐流都换成自己的,0成本打造专属于你的全网内容发现Agent-OpenBiliClaw

本帖使用社区开源推广,符合推广要求。我申明并遵循社区要求的以下内容: 我的帖子已经打上 开源推广 标签: 是 我的开源项目完整开源,无未开源部分: 是 我的开源项目已链接认可 LINUX DO 社区: 是 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是 以上选择我承诺是永久有效的,接受社区和佬友监督: 是 自从上两个帖子发表之后 【开源】受够 B 站"猜你喜欢”,我自己用新思路写了一个私有的bilibili推荐系统开源平替 ,【开源推广】打破信息茧房,用OpenBiliClaw养成专属于你的全网内容发现Agent ,有越来越多的用户来使用Openbiliclaw,也得到很多大佬的支