Python · ★ 61,831 · 🍴 9,037 · 📈 1,737 stars today
利用AI大模型，一键生成高清短视频 Generate short videos with one click using AI LLM.
中文介绍 基于 AI 大模型的一键短视频生成工具，用户只需提供主题或关键词，系统自动完成文案撰写、语音合成、字幕生成和视频拼接，输出高清短视频。适合内容创作者、自媒体运营者快速批量生产视频素材，大幅降低短视频制作门槛。
TypeScript · ★ 39,672 · 🍴 3,157 · 📈 4,466 stars today
Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
中文介绍 将任意代码库转换为可交互知识图谱的工具，支持探索节点、全文搜索并直接对代码提问。与 Claude Code、Codex、Cursor、Copilot、Gemini CLI 等主流 AI 编程工具集成，帮助开发者快速理解陌生代码库的结构与逻辑关系，提升代码审查和上手效率。
★ 5,659 · 🍴 434 · 📈 664 stars today
A skill file for removing AI tells from prose
中文介绍 专为 AI 写作润色设计的 skill 文件，用于识别并消除 AI 生成文本中的典型痕迹（如套话、生硬措辞、重复结构），让输出文字更接近自然人类写作风格。适合需要用 AI 辅助写作但又希望内容更具真实感的作者和编辑。
JavaScript · ★ 195,980 · 🍴 30,147 · 📈 2,062 stars today
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
中文介绍 面向 AI 编程 Agent 的性能优化框架，提供技能管理、本能响应、记忆系统和安全策略等模块，强调以研究驱动开发。兼容 Claude Code、Codex、Opencode、Cursor 等主流平台，帮助开发者构建更高效、更安全的自主编程 Agent 工作流。
Python · ★ 17,244 · 🍴 2,015 · 📈 695 stars today
Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork
中文介绍 Anthropic 官方开源的 Claude 插件集合，面向知识工作者在 Claude Cowork 环境中使用。涵盖文档处理、信息检索、工作流自动化等场景，通过插件扩展 Claude 的业务处理能力，适合企业和专业人员将 AI 深度融入日常知识型工作流程。
Shell · ★ 24,138 · 🍴 1,861 · 📈 2,715 stars today
Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop
中文介绍 为 AI 写作注入审美品位的 skill 配置文件，通过约束生成行为来避免千篇一律的平庸输出，帮助 AI 写出更有个性、更具表现力的文字。适合对文风有较高要求、不满足于 AI 默认平淡输出的写作者和内容创作者使用。
Python · ★ 21,996 · 🍴 2,348 · 📈 219 stars today
Fully automatic censorship removal for language models
中文介绍 全自动移除语言模型内容审查限制的工具，通过技术手段绕过大模型的内容过滤机制，使其能够回应被屏蔽的话题或请求。面向研究人员和需要探索模型边界的技术用户，属于模型对齐与安全研究领域的实验性工具。
Python · ★ 26,863 · 🍴 4,655 · 📈 402 stars today
Kronos: A Foundation Model for the Language of Financial Markets
中文介绍 面向金融市场的基础大模型 Kronos，将金融市场数据视为一种语言进行建模与理解，支持价格走势预测、市场信号分析等任务。适合量化研究员、金融科技团队将大模型能力引入市场分析和交易策略开发场景。
Python · ★ 10,924 · 🍴 1,250 · 📈 885 stars today
754 structured cybersecurity skills for AI agents · Mapped to 5 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND & NIST AI RMF · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 26 security domains · Apache 2.0
中文介绍 为 AI Agent 提供 754 项结构化网络安全技能的知识库，覆盖 MITRE ATT&CK、NIST CSF 2.0、MITRE ATLAS、D3FEND 和 NIST AI RMF 五大安全框架。符合 agentskills.io 标准，可与 Claude Code、GitHub Copilot 等工具集成，助力安全 AI Agent 的构建与评估。
TypeScript · ★ 47,311 · 🍴 6,716 · 📈 520 stars today
The open alternative to Salesforce, designed for AI.
中文介绍 Salesforce 的开源替代品，专为 AI 时代设计的 CRM 平台。提供客户关系管理的核心功能，同时深度整合 AI 能力，支持自定义扩展。适合希望避免高额 SaaS 授权费用、同时需要灵活定制和 AI 增强能力的企业和销售团队使用。
Shell · ★ 1,804 · 🍴 197 · 📈 143 stars today
Claude Code Dedicated Development Harness - Achieving High-Quality Development Through an Autonomous Plan→Work→Review Cycle
中文介绍 专为 Claude Code 打造的自主开发框架，通过「计划→执行→评审」的自动循环驱动高质量代码产出。将 AI 编程过程结构化为可重复的工作流，减少人工干预，适合希望利用 Claude Code 完成复杂软件开发任务、追求开发质量一致性的工程师和团队。
HTML · ★ 169,080 · 🍴 3,194 · 📈 2,223 stars today
DigitalPlat FreeDomain: Free Domain For Everyone
中文介绍 DigitalPlat 推出的免费域名申请平台，面向个人开发者和学生等有域名需求但预算有限的用户。提供免费域名注册服务，降低建站门槛，适合个人项目、学习实践、开源展示等轻量级建站场景使用。
Shell · ★ 209,481 · 🍴 18,667 · 📈 1,680 stars today
An agentic skills framework & software development methodology that works.
中文介绍 一套实用的 Agent 技能框架与软件开发方法论，通过标准化技能模块和开发流程，提升 AI 驱动软件开发的可靠性与效率。面向希望将 AI Agent 真正落地于实际工程项目的开发者，提供经过验证、可直接使用的工程实践方案。
★ 46,510 · 🍴 4,875 · 📈 1,133 stars today
An advanced guide to learn English which might benefit you a lot 🎉 . 离谱的英语学习指南/英语学习教程/英语学习/学英语
中文介绍 一份专为中文母语者设计的进阶英语学习指南，系统覆盖听说读写各技能的提升方法、资源推荐和学习路径规划。内容务实接地气，适合已具备基础英语能力、希望突破瓶颈、向流利表达进阶的学习者参考使用。
Rust · ★ 16,837 · 🍴 1,104 · 📈 427 stars today
Effortlessly compose, extend, and observe every service in real-time for the first time ever.
中文介绍 面向服务编排与可观测性的开发平台，支持无缝组合、扩展各类服务，并提供实时监控能力。解决微服务架构中服务编排复杂、运行状态难以实时感知的痛点，适合需要快速构建和管理多服务系统的后端开发者和平台工程团队。
JavaScript · ★ 5,856 · 🍴 293 · 📈 728 stars today
Curated list of the best free apps for PC and mobile
中文介绍 精心整理的 PC 端与移动端免费应用推荐清单，覆盖生产力、创作、开发、娱乐等多个类别。为用户提供高质量免费软件的一站式索引，帮助普通用户和开发者快速找到满足需求的优质免费工具，避免为功能相近的付费软件买单。
TypeScript · ★ 40,200 · 🍴 4,046 · 📈 56 stars today
💖🧸 Self hosted, you-owned Grok Companion, a container of souls of waifu, cyber livings to bring them into our worlds, wishing to achieve Neuro-sama's altitude. Capable of realtime voice chat, Minecraft, Factorio playing. Web / macOS / Windows supported.
中文介绍 可自托管的 AI 虚拟伴侣框架，以 Neuro-sama 为目标，支持实时语音对话、Minecraft 游戏联动等交互场景。强调数据自主权，所有数据存储在用户本地，面向希望构建或定制专属 AI 虚拟角色（Vtuber/陪伴型 AI）的开发者和爱好者。
👍 10
Social deduction games have become a popular testbed for probing reasoning, deception, coordination, and belief modeling in Large Language Model (LLM) agents. However, most environments are scored only by game outcomes such as win rates and largely remain to text-only interaction, making it difficul
👍 6
We introduce Gemini Embedding 2, a native multimodal embedding model that allows embedding video, audio, image, and text modalities in a unified representation space. We leverage the multimodal capabilities of Gemini to produce embeddings for arbitrary combinations of interleaved inputs across all t
👍 4
Recent advances in large language models (LLMs) have facilitated the widespread deployment of LLMs as interactive agents capable of reasoning, planning, and tool use. Despite strong performance on existing benchmarks, such agents often exhibit notable degradation when deployed in real-world settings
👍 9
Large language models (LLMs) have evolved into interactive agents that collaborate with users in real-world tasks. Effective collaboration in such settings increasingly depends on understanding the user beyond what is explicitly stated, as user intent is often reflected in fragmented daily interacti
👍 21
Test-Time Scaling (TTS) enhances the reasoning capabilities of large language models by allocating additional inference compute to explore the solution space. However, existing parallel TTS methods typically keep branches isolated during search: intermediate discoveries remain branch-private and can
👍 6
Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billio
👍 7
Large Language Models (LLMs) are increasingly deployed as autonomous agents that reason, use tools, and act over multiple steps. Yet most hallucination benchmarks still evaluate only the final output, missing failures that originate in intermediate Thought-Action-Observation steps. We present Trajel
👍 3
Layered image generation and editing is a fundamental capability that enables layer-wise reuse, editing, and composition of generated visual content, analogous to word-level editing in natural language. Despite its importance, this remains an underexplored area at scale. To address this gap, we pres
👍 11
Agentic reinforcement learning (RL) has proven effective for training LLM-based agents with external tool-use capabilities. However, we identify that agentic RL training induces increasing redundant tool calls and blurs the model's intrinsic knowledge boundary, where the model fails to distinguish w
👍 25
We introduce the MiniMax-M2 series, a family of Mixture-of-Experts language models built around the principle that mini activations can unleash maximum real-world intelligence. The flagship M2 contains 229.9B total parameters with only 9.8B activated per token. Designed end-to-end for agentic deploy
👍 12
Normalization layers in modern large language models (LLMs) consist of a deterministic normalization operation and a learnable scale vector. While the normalization operation has been extensively studied, the scale vector remains poorly understood despite its ubiquitous use. In this work, we present
👍 91
Vision-language models (VLMs) commonly formulate visual grounding and detection as a coordinate-token generation problem, serializing each 2D box into multiple 1D tokens that are learned and decoded largely independently. This token-by-token decoding mismatches the coupled structure of box geometry
👍 10
Large language model (LLM) agents rely on reusable skills to solve complex tasks. However, existing skill creation approaches treat skills as isolated and static artifacts, limiting their reusability, reliability, and long-term improvement. We propose MUSE-Autoskill Agent (Memory-Utilizing Skill Evo
👍 57
While spatial foundation models have demonstrated impressive performance on standard datasets, a critical question remains: are they truly all-round players capable of generalizing robustly across diverse downstream tasks, arbitrary viewpoints, shifting scene domains, varying input densities, and sp
👍 1
Can large language models detect and report their own internal states? A number of studies have argued that the answer to this question is yes. We argue, based on lessons from human metacognition research, that this conclusion may be premature: to be convinced of this conclusion we need to distingui
👍 1
Prior work establishes that controlled contrastiveness between self-generated responses from large language models, set via reward scores, improves downstream preference tuning in English. We extend this method to multiple languages and evaluate two models across a total of 14 high and low-resource
👍 31
Despite the emergence of diffusion large language models (D-LLMs) as an alternative to autoregressive large language models (AR-LLMs), safety monitoring for D-LLMs remains largely unexplored. Unlike AR-LLMs, D-LLMs generate text through a multi-step denoising process, exposing intermediate hidden re
👍 17
We introduce LLaVA-OneVision-2 (LLaVA-OV-2), the most capable vision-language model in the LLaVA-OneVision series to date, achieving superior performance across a broad range of multimodal benchmarks. The model builds on a native OneVision-Encoder and incorporates Windowed Attention for efficient lo
👍 48
We present MobileGym, a browser-hosted, lightweight, fully controllable environment for everyday mobile use, targeting interaction fidelity without replicating proprietary backends. It enables two capabilities previously out of reach for everyday apps: verifiable outcome signals through deterministi
👍 32
Audio-visual generation is rapidly advancing from short clips to minute-long content, while existing evaluation protocols remain largely confined to short-form settings. Existing benchmarks primarily focus on 5--10 second text-conditioned generation and rarely support unified evaluation across text,
👍 6
Subject-driven image generation aims to synthesize new images that preserve the identity of the given subject while following textual instructions. Existing approaches often encode text and reference images separately. This limits cross-modal reasoning abilities and causes copy-paste artifacts. Rece
👍 10
Visual reasoning through reinforcement learning with verifiable rewards (RLVR) has achieved remarkable progress. However, when dealing with multi-source inputs, existing approaches tend to treat them as a mere accumulation of information, lacking explicit mechanisms to distinguish whether integratin
👍 0
Large language models for vertical domains are bottlenecked by the scarcity of complex, domain-specific task-oriented dialogues. Existing data acquisition pipelines face a persistent trilemma: expert annotation is expensive, real-world service conversations are constrained by privacy and commercial
👍 7
Multi-agent LLM systems improve reasoning by combining outputs from multiple agents, but interaction-heavy methods can introduce error propagation and high communication overhead. When agents exchange raw responses or reasoning traces, incorrect intermediate reasoning may be adopted and amplified, l
👍 3
Long-horizon agentic reasoning requires large language models to act over long interaction histories containing thoughts, tool calls, observations, and partial conclusions. The challenge is not merely that these histories grow long, but that information needed for the current decision may be scatter
👍 4
LLM-based agents for GPU kernel generation are advancing rapidly, yet their progress is fundamentally constrained by the benchmarks they optimize against. Existing benchmarks are poorly aligned with production inference frameworks: they evaluate kernels on a single GPU with synthetic inputs, ignore
👍 72
The rapid evolution of generative video foundation models has propelled the field toward professional-grade cinematic synthesis. To achieve such demanding quality, the community transitions towards Reinforcement Learning (RL) and agentic workflows. However, reliable evaluation has emerged as a criti
👍 2
Agentic systems are becoming more capable: agents define strategies, take actions, and interact with different environments. This autonomy poses serious challenges for overseeing and assessing agent behavior. Most current tools are limited, focusing on observability with basic evaluation capabilitie
👍 2
Large language models (LLMs) are typically trained on shuffled corpora, yielding models whose knowledge is frozen at train time and whose temporal grounding remains poorly understood. In this work, we study the impact of pre-training dynamics on the acquisition of time-sensitive factual knowledge, f
👍 21
We study series-level cinematic remaking, a long-horizon video-to-video generation problem that localizes full episodes or films via stylization or actor replacement while strictly preserving narrative structure, motion choreography, and character identity across hundreds of shots. Existing video ge
@sairahul1 · 106.0K 粉丝 · 2.7M 阅 · 1.4K 赞 · 196 转
I thought I was using AI to code. I was actually just typing faster. Here is the difference — and the 7-agent system that changed everything. Save this. It will save you months. THE PROBLEM NOBODY
中文介绍 分享一套基于 Claude Code 的 7 个 Agent 协作系统，核心思路是从「用 AI 加速敲代码」升级为「让 AI 自动交付功能」。作者认为多数人只是用 AI 打字更快，而非真正自动化开发流程。该系统可在无人值守时持续交付特性，适合想构建软件工厂级 AI 工作流的开发者。
@cyrilXBT · 181.0K 粉丝 · 127.8K 阅 · 533 赞 · 80 转
Most people start their day the same way. They open Twitter and spend 20 minutes scrolling through noise looking for the three things that actually matter. They open their email and get pulled into
中文介绍 介绍如何用 Claude 搭建一个每日晨间信息简报 Agent：自动抓取互联网内容，在 5 分钟内生成当日关键摘要，替代每天刷 Twitter 或处理邮件噪音的低效习惯。定位为个人信息流自动化工作流教程，适合想用 AI 做信息过滤的用户。
@poteto · 26.6K 粉丝 · 86.5K 阅 · 540 赞 · 48 转
I need to get something off my chest. Before my interview @cursor_ai, I had never actually used Cursor. At Meta, Claude Code was explosively taking off. I even paid for a personal $200 a month plan
中文介绍 作者坦承入职 Cursor 面试前从未实际用过 Cursor，此前在 Meta 主要使用月费 200 美元的 Claude Code，并对其快速普及印象深刻。帖子分享了加入 Cursor 后的真实使用心得与工作流对比，视角较罕见——来自一位深度 Claude Code 用户转型为 Cursor 内部人员的切身体会。
@TheAhmadOsman · 59.9K 粉丝 · 54.5K 阅 · 512 赞 · 65 转
At some point, reading about LLMs stops being enough. You need to build the stack yourself: Tokenizer first, then embeddings, position, attention, Transformer blocks, objectives, decoding, cache, long
中文介绍 2026 版 LLM 工程实践路线图，强调光读理论不够，需要动手从零构建完整技术栈：依次覆盖 Tokenizer、Embedding、位置编码、Attention、Transformer 块、训练目标、解码策略、KV Cache 及长上下文处理等模块，是一份系统性的 LLM 工程项目学习指南。
@difflawb · 20.3K 粉丝 · 21.9K 阅 · 1.1K 赞 · 389 转
How a 40-line shell script became infrastructure In August 2024, Andrej Karpathy — co-founder of OpenAI, former AI Director at Tesla — published something unexpectedly small. Not a paper. Not a model.
中文介绍 复盘 Andrej Karpathy 于 2024 年 8 月发布的一个 40 行 Shell 脚本如何演变为重要基础设施。作者以此为切入点，探讨极简代码在 AI 工程生态中的影响力，属于对一段技术传播史的回溯分析，而非新工具教程。
@ActionModelAI · 57.1K 粉丝 · 5.8K 阅 · 505 赞 · 344 转
We are witnessing the beginning of the biggest economic shift in modern history. And most people still don’t realize it. AI replacement is no longer some distant sci-fi prediction. It has started.
中文介绍 观点类帖子，宣称 AI 替代就业已从预测变为现实，是「现代史上最大经济转型」的起点。内容以危机感叙事为主，缺乏具体数据或工具细节，整体属于 AI 冲击劳动力市场的泛泛论述，信息密度较低。
Datasets vs. inductive bias, world models, and programmable biology
中文介绍 Meta AI研究员Alex Rives探讨ESMFold2蛋白质结构预测模型的进展，围绕数据集与归纳偏置的权衡、世界模型构建以及可编程生物学展开讨论，认为"苦涩教训"（即大规模数据与算力优于人工先验知识）正在蛋白质领域得到验证。
中文介绍 Artificial Analysis与IBM联合发布ITBench-AA基准测试，专门评估AI代理在企业IT任务中的表现。测试结果显示，当前前沿模型得分均低于50%，表明现有大模型在处理真实企业IT自动化任务方面仍存在显著差距。
Cisco and OpenAI are redefining enterprise engineering with Codex, helping Cisco scale AI-native development, accelerate AI Defense work, and automate defect remediation.
中文介绍 思科（Cisco）与OpenAI宣布合作，共同利用Codex重塑企业级软件工程实践。合作内容涵盖扩展AI原生开发能力、加速AI Defense安全产品研发，以及实现缺陷自动修复，旨在提升思科整体软件开发效率与质量。
See how OpenAI, Thrive, and Crete built a self-improving tax agent with Codex, automating filings, improving accuracy, and accelerating workflows.
中文介绍 OpenAI联合Thrive和Crete，基于Codex构建了一套自我改进型税务代理系统，可自动完成税务申报、提升申报准确率并加快工作流处理速度。该案例展示了大型语言模型在专业财税场景中实现自动化与持续优化的实际应用路径。
it's funding news, but it's good news.
中文介绍 AI基础设施领域迎来新一批百亿美元级（十角兽）企业，Fireworks AI和Baseten相继完成大额融资，OpenRouter也被报道即将跟进融资。上述动态反映出市场对AI推理与模型部署基础设施的持续强劲需求。
Warp uses GPT-5.5 and OpenAI models to coordinate coding agents across local, cloud, and open-source development workflows.
中文介绍 终端工具Warp宣布采用GPT-5.5及OpenAI系列模型，用于协调跨本地、云端及开源环境的编程代理工作流。Warp押注GPT-5.5的能力，以支持更复杂的多代理协作编程场景，并加大对开源开发模式的投入。
Ahead of global elections, we’re helping people access information, supporting cyber defenders, and increasing AI transparency
中文介绍 面向2026年全球选举周期，OpenAI宣布推出系列选举信息保障措施，包括帮助用户获取权威选举资讯、支持网络安全防御工作，以及提升AI生成内容的透明度，以应对AI可能对选举信息环境造成的潜在干扰。
中文介绍 Hugging Face发布博客介绍Reachy Mini机器人实现完全本地化运行的方案，使该小型机器人的对话与交互能力无需依赖云端服务，可在本地设备上独立完成推理与响应，进一步降低机器人智能化的部署门槛。
中文介绍 Hugging Face在TRL框架中推出Delta Weight Sync（增量权重同步）功能，支持通过Hub存储桶高效传输万亿级参数模型的增量权重，大幅降低大规模模型训练与微调过程中的参数同步带宽消耗与存储成本。
中文介绍 本期TLDR AI要闻涵盖：xAI旗下产品对Cursor使用设限；DeepSWE软件工程代理模型发布；中国针对AI相关人员出行出现限制性措施。上述事件折射出AI工具生态竞争加剧及地缘科技博弈持续升温的趋势。
Amid rapidly growing adoption of enterprise-level AI agents, there’s a disconnect emerging between ambition and execution. Although 85% of organizations say they want to be agentic within the next three years, 76% say their current operations and infrastructure can’t support that change. They cite a
中文介绍 《麻省理工科技评论》指出，企业在落地AI代理方面存在明显落差：85%的组织希望在未来三年内实现"代理化"运营，但76%承认现有基础设施无法支撑这一目标。报告呼吁企业重新审视组织架构设计，以匹配AI代理时代的运营需求。
Haven’t you heard? White-collar jobs are going away, decimated by AI. Waves of layoffs in the tech sector (most recently at Coinbase and Meta and Cisco) are said to presage what will soon come for all of us knowledge workers. But before you quit your job as a software developer or financial analyst—
中文介绍 《麻省理工科技评论》对"AI大规模取代白领"的舆论热潮进行理性审视。文章指出，尽管Coinbase、Meta、Cisco等科技公司近期裁员频发，但目前尚无充分证据表明AI已从整体上造成大规模职业消失，提醒公众避免过度恐慌。
Artificial intelligence has not so far produced a clean story of mass unemployment. Aggregate employment in developed countries remains broadly stable, and recent assessments have found limited evidence that AI has shifted the headline numbers. But a troubling change may be hiding beneath the surfac
**Harness engineering** is emerging as the key differentiator for coding agents, emphasizing the stack of **model + harness + eval loop** over just stronger base models. **DeepSeek** is building a harness team to optimize interaction and verification loops, while **Google's Gemini Managed Agents** a
中文介绍 编程代理领域，"工具链工程"（Harness Engineering）正成为关键差异化因素，核心在于"模型+工具链+评估循环"的整体协同，而非单纯追求更强基础模型。DeepSeek正在组建专属工具链团队，Google Gemini等也在优化交互与验证流程。
中文介绍 本期TLDR AI要闻涵盖：xAI发布Grok命令行工具Grok Build CLI；AI硬件市场最新动态；教皇利奥就人工智能风险发出警示，呼吁关注AI对人类尊严与社会伦理的潜在影响。
从本期21只标的的技术面整体来看，各资产类别呈现明显分化格局。 美股方面，SPY 和 QQQ 均处于多头排列，RSI14 分别为 71.1 和 74.6，双双进入超买区间，SPY 距52周高点仅 -0.22%，QQQ 距52周高仅 -0.53%，指数层面强势延续。个股中 AAPL 的 RSI14 高达 78.8 为全场最高，MACD 金叉强化上行动量；TSLA 5日涨幅 +8.97%，MACD 金叉刚刚触发，META 当日大涨 +3.74%并出现 MACD 金叉，中期均线有所修复。相对弱势的是 MSFT，RSI14 仅 49.6，价格（412.67）低于 SMA200（459.33），趋势判定为中性。NVDA 出现 MACD 死叉，近5日下行 -3.63%，动量有所钝化。宏观指标方面，VIX 当前报 16.29，5日下行 -6.59%，低于 SMA20（17.39）和 SMA50（20.31），市场隐含波动率整体处于低位，情绪相对平稳。10年期美债收益率维持在 4.48%，多头排列但 MACD 死叉出现，美元指数 DXY（99.22）多头排列并接近52周高点。 加密市场整体承压显著。加密恐慌贪婪指数降至 25（极度恐慌），总市值24小时下跌 -1.79%至 2.57 万亿美元，BTC 主导率 57.8%。BTC、ETH、SOL 三者均形成空头排列，RSI14 分别为 37.6、31.7、39.0，价格全线位于 SMA20/50/200 之下，ETH 距52周高点回落幅度达 -59.25%，技术面弱势明显。 中概股方面，阿里、拼多多、京东、腾讯均呈空头排列，PDD 单日暴跌 -10.38%，腾讯控股接近52周低点，RSI14 报 32.3，整体板块技术面偏弱。商品方面，黄金期货 RSI14 为 37.2，价格（4480.90）跌破三条均线；WTI 原油5日跌幅 -16.49%，MACD 死叉确认，是近期技术面最弱的商品标的之一。
PDD 当日跌幅达 -10.38%，5日累计跌幅 -11.02%，距52周高点已下探 -37.87%。当前价格 86.61 美元，已跌破 SMA20（97.30）和 SMA50（99.17），形成空头排列。RSI14 仅 32.7，接近超卖区间。MACD 死叉且柱状图负值持续扩大，信号线与 MACD 线均处于负值区域（-1.8143 vs -1.1408），下行动量显著。
QQQ 当前报价 729.45，位于 SMA20（701.81）、SMA50（647.42）、SMA200（616.24）之上，呈多头排列。5日涨幅 +3.98%，距52周低点上涨 +42.49%，距52周高仅 -0.53%。RSI14 报 74.6，进入超买区间，但 MACD（20.96）高于信号线（21.47），差距收窄，柱状图略有收缩，短期动量稍有减弱迹象，需关注是否出现回调信号。
BTC 当前报价 74,194.55 美元，低于 SMA20（78,264.78）、SMA50（77,121.87）和 SMA200（80,127.51），空头排列成立。1日跌幅 -2.15%，5日 -1.71%，距52周高点 -41.21%。RSI14 为 37.6，逼近超卖边界。MACD 深度负值 -542.85，信号线 -10.95，MACD 向下背离信号线幅度悬殊。结合加密市场恐慌贪婪指数 25（极度恐慌）及总市值24小时下跌 -1.79%，链上情绪持续低迷。
AAPL 当前报价 310.85 美元，当日上涨 +0.82%，5日 +3.97%，距52周高仅 -0.77%。SMA20（293.39）、SMA50（272.88）、SMA200（262.40）多头排列，价格全线位于三条均线之上。RSI14 达 78.8，处于明显超买状态，为21只标的中 RSI 最高值。MACD（10.27）上穿信号线（9.42），柱状图翻正，短期上行动量仍在持续。
WTI 原油当前报价 90.00 美元，1日跌幅 -4.14%，5日累计跌幅高达 -16.49%，为所有候选标的中5日跌幅最大。价格已跌破 SMA20（100.36）和 SMA50（98.22），RSI14 为 40.2，接近超卖区间。MACD（-0.28）下穿信号线（1.19），死叉确认，下行动量正在加速。虽然 SMA200（71.80）仍在价格之下提供远端支撑，但短中期技术面弱势明显。
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所得技术指标进行客观描述，仅供技术指标解读参考，不构成任何投资建议或买卖信号。过去走势不代表未来表现，市场存在不确定性，读者应结合自身风险承受能力独立决策。
Bundaberg man, 36, allegedly drove into and assaulted multiple people in Childers and North Isis on Wednesday A man has been charged with murder after allegedly beating a 78-year-old pedestrian to death with a metal pole during a violent and random rampage that also injured two other men. A 36-year-
中文摘要 澳大利亚昆士兰州一名36岁男子被控谋杀罪。据称他周三在奇尔德斯及北艾西斯地区驾车冲撞并袭击多名路人，其中一名78岁老人遭其用金属杆殴打致死，另有两人受伤。案件被描述为随机暴力袭击事件。
Financial services minister says it would mean banks and telcos would focus dispute resolution processes on bigger scams. Follow updates live Get our breaking news email, free app or daily news podcast Scams come in all shapes and sizes (and none of them are nice), and the government is considering
中文摘要 澳大利亚工党推出新计划，拟自动向诈骗受害者赔付3000澳元。金融服务部长表示，该措施将促使银行和电信公司将纠纷解决资源集中于较大规模的诈骗案件，以提升对受害者的保护力度。
President Trump said U.S. politics would not affect his strategy in the unpopular war and was open to protracted peace talks. He also insisted that Iran would not control the Strait of Hormuz.
中文摘要 美伊战争谈判持续胶着，特朗普表示美国国内政治不会影响其对伊策略，并称不在意中期选举压力。他同时表明愿意接受旷日持久的和平谈判，并坚持伊朗不得控制霍尔木兹海峡。
Exclusive: Robinson, who has been at Reuters for 16 years, will be responsible for about 2,000 staff across the ABC’s news division A top news executive from Reuters, Simon Robinson, is expected to be appointed as news director of the ABC in coming days, Guardian Australia can reveal. He will be the
中文摘要 路透社高管西蒙·罗宾逊预计将在数日内被任命为澳大利亚广播公司（ABC）新闻总监，接替贾斯汀·史蒂文斯。罗宾逊在路透社任职已逾16年，新职位将负责管理ABC新闻部门约2000名员工。
President Trump said there would be serious repercussions if Iran and Oman move forward with a deal to open the Strait of Hormuz.
中文摘要 特朗普警告阿曼，若其与伊朗推进开放霍尔木兹海峡的协议将面临严重后果，但随后语气软化称「他们没事的」。此举显示美国在霍尔木兹海峡控制权问题上对地区国家施加外交压力。
The government is letting people connect with the world after a near-total internet shutdown. But not everyone has access, and those who do wonder how long it will last.
中文摘要 伊朗在实施近全面断网封锁88天后，开始允许民众重新连接互联网。然而并非所有人均可访问，已恢复网络的民众亦对管控措施能否持续解除持怀疑态度。此前断网期间，电视、通讯及新闻均受到严格审查。
Seven confirmed cases of the virus have already been reported in Kampala, the capital, but officials say the country has robust disease surveillance.
中文摘要 乌干达因担忧埃博拉疫情扩散而关闭与刚果的边境。目前首都坎帕拉已确认7例感染病例，但官员表示乌干达建有完善的疾病监测体系。疫情蔓延风险引发国际社会高度关注。
The BBC visited China’s EV factories and found they are dominating the ecosystems shaping the global auto industry.
中文摘要 BBC深入中国电动汽车工厂调查发现，中国车企正主导全球汽车产业生态链，全球各大传统汽车制造商在技术、成本及供应链方面均面临来自中国电动车企业的强大竞争压力，处境日趋艰难。
Santiago Rivera is widely credited with creating the “burnt” cheesecake in the 1980s, though he doesn’t love the spinoffs it has spawned. Decades later, he’s preparing to hand over his kitchen to his children.
中文摘要 被誉为「巴斯克烤芝士蛋糕」发明者的西班牙人圣地亚哥·里维拉宣布即将退休，计划将厨房交棒给子女。他于1980年代创作出这款「烤焦」芝士蛋糕，但对于由此衍生出的众多仿制品，他本人并不热衷，且坦言个人更偏爱巧克力。
In Nepal, a nation wedged between India and China, a unique time zone is just one expression of a singular national identity.
中文摘要 尼泊尔采用与周边国家相差15分钟的独特时区，这一特殊设置不仅是其夹处印中两国之间的地理现实体现，也是尼泊尔独特国家认同的一种表达。该时区安排折射出尼泊尔在地区格局中彰显独立性的历史选择。
Markets betting a deal will reopen the Strait of Hormuz and soothe the deep global economic uncertainty.
中文摘要 受美伊可能达成协议的预期提振，全球金融市场普遍上涨。投资者押注协议将重新开放霍尔木兹海峡，从而缓解全球经济不确定性。市场情绪因和平谈判进展而明显改善。
A truce isn’t necessarily doomed if the missiles are still flying. Here’s why.
中文摘要 《纽约时报》分析指出，停火协议与真正意义上的和平之间存在本质差异。即便导弹仍在飞行，停火也未必注定失败。文章以黎以及美伊局势为背景，探讨停火协议的实际效力与局限性。
Sentencing of Kenneth Iwamasa concludes prosecution of five people in connection with Perry's death from drug ketamine.
中文摘要 美国电视明星马修·派瑞的前助理肯尼斯·岩政被判处41个月监禁。此案为涉及派瑞因氯胺酮用药死亡的相关起诉案之一，共5人被追诉。随着此次宣判，检察机关对本案的起诉工作宣告结束。
Intensification comes while US-Iran indirect talks continue.
中文摘要 尽管停火协议仍在名义上维持，以色列近期仍持续加大对黎巴嫩的军事打击力度。分析指出，此轮升级恰逢美伊间接谈判进行之际，外界关注以色列此举对地区停火框架及美伊谈判进程的潜在影响。
Donald Trump says he won’t allow Iran to control the Strait of Hormuz under the terms of any deal reached with Tehran.
中文摘要 特朗普明确表态，在任何与伊朗达成的协议框架下，美国都不会允许伊朗控制霍尔木兹海峡。此番表态被视为美国在伊核及地区安全谈判中坚守的核心红线之一。
Gold held a two-day drop, with a lack of progress in US-Iran peace talks continuing to pressure interest rates.
中文摘要 黄金延续两日跌势，美伊和谈未取得实质进展，市场对通胀风险的担忧持续对利率形成压力，金价承压维持低位。
With US waiting staff getting cross at receiving less than 20%, tips are also on the rise elsewhere.
中文摘要 美国服务业小费文化正向海外蔓延。美国餐饮从业者对低于20%的小费比例表示不满，与此同时其他国家和地区的小费金额也呈上升趋势，引发外界对小费文化全球扩散的讨论。
By launching a gold-clearing system in the next couple of months, Hong Kong is set to secure first-mover advantage in a push to become Asia’s preeminent hub for bullion trading.
中文摘要 香港计划在未来数月内推出黄金清算系统，以争取成为亚洲最重要黄金交易中心的先发优势。此举是香港强化黄金枢纽地位战略的重要组成部分。
Synopsys Inc. said it has reached an agreement with Elliott Investment Management and is appointing the activist investor’s Jesse Cohn to the chip-design software maker’s board.
中文摘要 芯片设计软件商Synopsys与激进投资者Elliott Investment Management达成和解，将任命Elliott合伙人Jesse Cohn加入公司董事会。
Asian stocks looked set for a mixed start after conflicting signals from the US and Iran over prospects for a deal to end the war. Crude oil gained.
中文摘要 受美伊谈判释放矛盾信号影响，亚洲股市开盘预计表现平淡。美伊双方就结束战争达成协议的前景不明朗，原油价格小幅上涨。
PacSun CEO Brieane Olson said that the retailer has built a resurgence by 'co-creating' and listening to a live feedback loop of what Gen Z consumers want to see from the retailer instead of treating young people like a static audience. She said brands have to shit the notion of control and relinqui
中文摘要 美国零售商PacSun首席执行官Brieane Olson表示，公司通过与Z世代消费者「共创」品牌的策略实现复苏，持续收集年轻消费者的实时反馈，而非将其视为静态受众，以此重塑品牌形象。
Ohio will pause incentives for future data center projects, following a directive from the governor to halt the tax credits while a committee studies the impact of the projects on the state’s economy.
中文摘要 美国俄亥俄州州长下令暂停针对新数据中心项目的税收优惠政策，同时成立委员会评估数据中心项目对州内经济的综合影响，此举源于各方反对声音日益增多。
Oil edged higher as the US and Iran remained at odds over how to reopen the Strait of Hormuz, with the war entering its fourth month.
中文摘要 国际油价小幅走高。美伊两国在重开霍尔木兹海峡问题上分歧持续，中东战争已进入第四个月，协议前景不明，供应风险支撑油价。
The BBC visited China’s EV factories and found they are dominating the ecosystems shaping the global auto industry.
中文摘要 BBC记者走访中国电动车工厂后发现，中国车企正主导全球汽车产业生态系统的重塑，全球主要汽车制造商在与中国同行的竞争中普遍面临压力。
Eric Weiner, Bloomberg News senior editor for equities Americas and Sarah Hunt, chief market strategist at Alpine Saxon Woods, break down muted market reaction to Iran peace negotiations as oil remains high with the future of MidEast shipping lanes uncertain. (Source: Bloomberg)
中文摘要 彭博新闻高级编辑Eric Weiner与Alpine Saxon Woods首席市场策略师Sarah Hunt分析指出，市场对美伊和谈进展反应平淡，中东航运通道前景不确定性持续，油价维持高位。
A major review examining the causes of rising youth unemployment says getting on the career ladder is now "out of reach" for many.
中文摘要 一份重要报告警告称，若不采取有效行动，未来五年内六分之一的年轻人将处于既未就业也未接受培训的状态。报告指出，对许多年轻人而言，踏上职业发展之路已愈发困难。
Also in today’s newsletter: Samsung workers set for $400,000 bonus, and Byju’s founder sentenced to jail in Singapore
中文摘要 《金融时报》今日要闻：多国领导人赴华会见习近平；三星员工有望获得约40万美元奖金；印度在线教育公司Byju's创始人在新加坡被判处监禁。
18 回复 · 程序员 节点
10 回复 · 程序员 节点
7 回复 · 程序员 节点
24 回复 · Apple 节点
7 回复 · Apple 节点
9 回复 · Apple 节点
18 回复 · Apple 节点
16 回复 · Apple 节点
10 回复 · Apple 节点
43 回复 · Apple 节点
2026 年主流终端 AI 编码工具分类 类型 工具 特点 最佳使用场景 TUI Grok Build 多面板 TUI + Plan Mode + Todo 管理 + 并行子代理，终端原生 复杂工程任务、重构、大型代码库 TUI OpenCode 主题化 TUI、分割面板、LSP 支持，接近 IDE 体验 可视化交互和多模型切换 TUI Crush (Charmbracelet) 美观直观 TUI，响应快，支持 LSP 代码理解 日常编码建议、快速迭代 TUI Crustly / Ralph-tui / Sidecar Rust 实现或专用仪表盘 TUI 本地优先、代理编排、监控 CLI Ai
享受吧，各位 25 个帖子 - 23 位参与者 阅读完整话题
Mac风格的web页面样式 完整的功能 佬们还有哪里可以改改的吗 1m小鸡 速度慢 点击查看 56 个帖子 - 46 位参与者 阅读完整话题
今年是我打工的第一年，也是在L站的第一年。（工作后体重在不断增加 ） 。 没有给蜡烛，吹的叔叔的打火机，我们家很正式，还放了海底捞生日歌。 我们家会给我们生日过到成家有孩子，然后再给孩子过生日。 （同理压岁钱也会给到有孩子，想到我每月才挣几个钱，还好还不用给压岁钱 ） 总结：今天很开心，祝佬友们今年也顺利呀！ 25 个帖子 - 21 位参与者 阅读完整话题
开源推广声明 （点击了解更多详细信息） openclaw 爆火之后各种 agent 产品如雨后春笋般涌现出来，但是聚焦端侧/手机上的 agent 却一直很少。我们从去年开始做手机gui自动化的场景，今年专注做 native agent。近期更新了很多，软件的稳定性都大幅度提升！ 强拓展性、工具性：文件格式转换（ffmpeg、pandoc）、文件重命名etc，你可以在手机上完成电脑系统可以做的大部分事情。 2、沙盒化：软件即agent系统，不会对手机系统造成任何损伤（养的正确方式）——>完全优于pc agent 你甚至可以让agent直接修改软件所存储的数据，自定义程度很高。 3、丰富上下文——
-----后续----- 65 个帖子 - 45 位参与者 阅读完整话题
13 个帖子 - 13 位参与者 阅读完整话题
要说互联网上最容易吵起来的话题，涉及某*、某*、某*、某*的话题绝对拥有姓名。我们发现不管是手机、汽车还是大模型，只要一发出来，很快就能吵起来。 大聪明冥思苦想良久，终于想到一个止纷争的好办法。现在我们宣布：L站已同时接受某*、某*、某*、某*的投资！ 所谓不看僧面看佛面，都客气点，再攻击、抹黑某*、某*、某*、某*，我们可就不客气了。 到底投没投，你别管；具体是哪些公司，你猜。 162 个帖子 - 152 位参与者 阅读完整话题
刚才看到的，没有抢到冰佬的订阅，然后应该是被其他好心佬抽奖或者送了一个，结果转手就卖，，， 52 个帖子 - 45 位参与者 阅读完整话题
测试目标: 编号 查询主题 预期相关区域 1 驱动 probe 失败后如何按相反顺序释放已经初始化的资源 drivers/base、drivers/net、drivers/pci 2 设备移除时如何取消定时器、延迟工作队列，避免卸载后的回调访问已释放对象 drivers/net、kernel/workqueue.c、include/linux/timer.h 3 使用引用计数管理内核对象生命周期，最后一个引用释放时执行清理 lib、include/linux/kref.h、kernel 4 使用 RCU 保护读路径，更新路径删除对象后延迟释放内存 kernel/rcu、include/linu