Python · ★ 11,917 · 🍴 1,538 · 📈 2,935 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、代理驱动的视频制作系统，包含12条管线、52种工具和500余项代理技能。能将AI编程助手转化为完整的视频制作工作室，适合内容创作者和开发者自动化视频生产。
Swift · ★ 7,291 · 🍴 501 · 📈 2,462 stars today
macOS video editor built for AI
中文介绍 专为AI打造的macOS视频编辑器，集成智能剪辑、自动生成等AI能力，帮助用户高效完成视频创作。适合内容创作者和视频编辑者使用AI辅助工作流。
TypeScript · ★ 32,191 · 🍴 3,930 · 📈 508 stars today
The open-source AI voice studio. Clone, dictate, create.
中文介绍 开源AI语音工作室，支持语音克隆、听写和创作。可用于语音合成、有声书制作、虚拟主播等场景，降低语音内容生成门槛。
Python · ★ 18,652 · 🍴 2,209 · 📈 957 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 为AI代理提供817个结构化网络安全技能，映射至MITRE ATT&CK、NIST CSF等6大主流框架。便于构建自动化安全分析和响应系统，适用于安全研究及SOC自动化。
Clojure · ★ 52,846 · 🍴 3,385 · 📈 730 stars today
Penpot: The open-source design tool for design and code collaboration
中文介绍 开源设计协作工具，面向设计师和开发者，支持实时协作、原型制作与代码导出，是Figma的开放替代方案。
TypeScript · ★ 82,862 · 🍴 7,237 · 📈 691 stars today
#1 PDF Application on GitHub that lets you edit PDFs on any device anywhere
中文介绍 GitHub上排名第一的PDF应用，提供在线或离线编辑、转换、合并、拆分等全面功能，支持任意设备随时随地处理PDF文件。
TypeScript · ★ 113,110 · 🍴 16,795 · 📈 649 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 Garry Tan的Claude Code精确配置，包含23个特定工具，模拟CEO、设计师、工程经理等角色，帮助开发团队利用AI高效管理开发流程。
TypeScript · ★ 29,955 · 🍴 2,821 · 📈 369 stars today
Write HTML. Render video. Built for agents.
中文介绍 基于HTML生成视频的开源框架，专为AI代理设计。开发者可用HTML描述场景，自动渲染为完整视频，适合动态视频内容自动化生产。
Rust · ★ 21,453 · 🍴 1,085 · 📈 538 stars today
Turso is an in-process SQL database, compatible with SQLite.
中文介绍 进程内SQL数据库，兼容SQLite但支持边缘计算和分布式部署，具备低延迟、高并发特性，适合移动端、IoT及Serverless应用。
Python · ★ 73,217 · 🍴 9,894 · 📈 736 stars today
An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.
中文介绍 字节跳动开源的长周期超级代理框架，具备研究、编码和创作能力。通过沙箱、记忆、工具、子代理等组件处理复杂任务，适用于自动化研发和多步工作流。
C · ★ 11,487 · 🍴 851 · 📈 1,186 stars today
High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static binary, zero dependencies.
中文介绍 高性能代码智能MCP服务器，将代码库索引为持久化知识图谱，支持158种语言，平均毫秒级查询，减少99% Token消耗。单个静态二进制文件，适合大规模代码库的快速理解与导航。
Python · ★ 45,775 · 🍴 41,902 · 📈 1,560 stars today
LLM 驱动的多市场股票智能分析系统：多源行情、实时新闻、决策看板与自动推送，支持零成本定时运行。 LLM-powered multi-market stock analysis system with multi-source market data, real-time news, decision dashboard, automated notifications, and cost-free scheduled runs.
中文介绍 LLM驱动的多市场股票智能分析系统，集成多源行情、实时新闻、决策看板与自动推送，可零成本定时运行。适合个人投资者进行自动化市场监控与分析。
TypeScript · ★ 137,235 · 🍴 7,962 · 📈 736 stars today
The API to search, scrape, and interact with the web at scale. 🔥
中文介绍 大规模网页搜索、抓取和交互API，支持结构化数据提取。适用于数据采集、内容聚合、AI训练数据准备等场景。
TypeScript · ★ 17,691 · 🍴 2,744 · 📈 63 stars today
Clone any website with one command using AI coding agents
中文介绍 利用AI编码代理，通过一条命令即可克隆任意网站。适合快速复制网站结构、前端模板或进行网页存档。
Jupyter Notebook · ★ 21,034 · 🍴 2,424 · 📈 187 stars today
AirLLM 70B inference with single 4GB GPU
中文介绍 单张4GB GPU即可运行70B参数大模型推理，通过双级模型压缩技术大幅降低显存需求，使低成本硬件也能部署大型语言模型。
Shell · ★ 141,573 · 🍴 12,251 · 📈 2,051 stars today
Skills for Real Engineers. Straight from my .claude directory.
中文介绍 来自知名工程师Matt Pocock的Claude技能集，包含面向真实工程师的CLI工具和AI助手配置，可提升日常开发效率。
👍 118
While 10B-level industrial foundation models have pushed the boundaries of image inpainting, their prohibitive computational costs severely hinder practical deployment. Constructing a highly optimized task-specific specialist offers a promising solution; however, extreme structural compression inevi
中文介绍 提出Moebius轻量级图像修复框架，0.2B参数即可达到10B级别性能。通过极端结构压缩和专门化设计，大幅降低计算成本，适用于实际部署。
👍 69
Dexterous interaction with articulated objects is important for household, assistive, and humanoid manipulation, where multi-finger hands can provide compliant contact patterns beyond parallel-jaw grasping. However, articulated-object manipulation differs from static-object manipulation: the target
中文介绍 提出DragMesh-2实现与铰接物体的灵巧手交互，考虑物理合理性。通过多指手提供柔顺接触，解决动态目标操作中的接触模式问题。
👍 56
LiveCodeBench (LCB) has recently become a widely adopted benchmark for evaluating large language models (LLMs) on code-generation tasks. By curating competitive programming problems, constantly adding fresh problems to the set, and filtering them by release dates, LCB provides contamination-aware ev
中文介绍 扩展LiveCodeBench至多编程语言（Multi-LCB），通过收集竞赛题目、持续更新并过滤污染，提供跨语言代码生成评估基准。
👍 50
Multimodal large language models (MLLMs) have achieved remarkable progress in visual understanding tasks. However, most existing MLLMs rely on autoregressive generation, which limits their efficiency for perception tasks that require captioning multiple regions. In this work, we propose PerceptionDL
中文介绍 提出PerceptionDLM，利用多模态扩散语言模型进行并行区域感知。克服自回归生成效率低的问题，实现多区域同时描述。
👍 44
Current agentic robot systems can write executable Code-as-Policy programs, observe feedback, and revise behavior across multiple attempts, but they remain largely task-driven: reusable skills are acquired only after explicit instructions. We study Playful Agentic Robot Learning, where an embodied c
中文介绍 研究玩耍式具身机器人学习，让机器人在无明确指令下自主探索并获取可复用技能，超越传统任务驱动模式。
👍 38
Real-world spatial intelligence requires reasoning over a continuous and evolving 3D world, yet existing VLMs and tool-augmented agents largely remain tied to static, stateless inference from isolated visual observations. We introduce \textsc{S-Agent}, a spatial tool-use agentic paradigm for underst
中文介绍 提出S-Agent空间工具使用智能体，通过与环境交互推理连续3D世界，突破现有VLM对静态孤立观察的局限。
👍 37
Agent benchmarks are growing fast, but no single benchmark touches more than four or five of the dimensions that deployment exposes. This paper aggregates the largest coordinated deep-dive of one MCP-based industrial-agent benchmark to date: fourteen parallel implementation studies covering new asse
中文介绍 超越静态排行榜，提出预测有效性评估框架，对MCP工业智能体基准进行14项并行实现研究，覆盖部署中多个维度。
👍 31
Advances in radiance fields have enabled photorealistic novel view synthesis. In several domains, large-scale real-world datasets have been developed to support comprehensive benchmarking and to facilitate progress beyond scene-specific reconstruction. However, for distractor-free radiance fields, a
中文介绍 发布DF3DV-1K大规模数据集与基准，用于无干扰物新视角合成。支持全面评估，推动超越场景特定重建的进展。
👍 26
Style-content dual-reference generation aims to synthesize an image that preserves the structure and semantics of a content reference while adopting the style of a separate style reference.Despite recent progress, this setting remains challenging because models must balance content fidelity, style a
中文介绍 提出FreeStyle风格内容双参考生成方法，从社区LoRA挖掘实现自由控制。平衡内容保真度与风格迁移，无需额外训练。
👍 14
Personalized presentation generation requires more than conditioning on a current prompt or template: agents must preserve stable user preferences across tasks, retain newly introduced preferences and constraints during multi-turn revision, and carry out local edits reliably. We propose MemSlides, a
中文介绍 提出MemSlides层次记忆驱动智能体框架，用于个性化幻灯片生成及多轮局部修订。保留用户偏好并支持可靠局部编辑。
👍 13
Memory benchmarks for LLM agents largely assume single-user settings, leaving shared assistants for hospitals, workplaces, campuses, and households understudied. In these deployments, multiple principals write to a common memory pool and query it under different roles, scopes, and relationships, so
中文介绍 提出GateMem基准，评估多主体共享记忆智能体的记忆治理能力。涉及医院、校园等场景，测试不同角色与关系下的记忆读写。
👍 11
Large language models (LLMs) often fail when answering requires identifying a small but decisive piece of evidence within a long or complex context, such as a single line in a tool trace or a subtle detail in an image. We propose ContextRL, a context-aware reinforcement learning (RL) method that imp
中文介绍 提出ContextRL上下文感知强化学习方法，提升LLM在长文本或复杂多模态情境下识别关键证据的能力，减少推理错误。
👍 11
Achieving dexterous robotic manipulation in the real world heavily relies on human supervision and algorithm engineering, which becomes a central bottleneck in the pursuit of general physical intelligence. Although emerging coding agents can generate code to automate algorithm search, their successe
中文介绍 提出ENPIRE框架，实现真实世界灵巧操作策略的自动自我改进。通过编码智能体自动搜索算法，减少人工干预。
👍 10
Multi-step LLM pipelines fail through interactions among retrieval, reasoning, and formatting steps, so prompt-only optimization can miss bottlenecks in the chain. We present FAPO (Fully Autonomous Prompt Optimization), a framework that lets Claude Code optimize an LLM pipeline inside a standardized
中文介绍 提出FAPO全自动提示优化框架，让Claude Code优化多步LLM管道中的检索、推理和格式化步骤，识别链式瓶颈。
👍 10
Visual thinking should not only sound right; it should show its evidence. While recent vision-language models (VLMs) can produce natural-language reasoning traces, these traces often leave the supporting image regions implicit, making them hard to verify and difficult to supervise. We introduce visu
中文介绍 引入视觉基础推理，要求VLM在生成文本推理时同时显示支撑图像区域。提高可验证性并便于监督。
👍 9
Retrieval-augmented generation (RAG) systems depend critically on how documents are chunked and searched. Fine-grained chunks can improve retrieval precision but expand the search space, increasing latency and cost; larger chunks reduce the number of candidates but make dense similarity less reliabl
中文介绍 提出MCompassRAG，利用主题元数据作为语义罗盘指导段落级检索。平衡细粒度与搜索空间，提升RAG精度和效率。
👍 9
While reasoning on autoregressive (AR) models is often performed by chain-of-thought reasoning and reflection, their refinement of previous outputs still relies on fully sequential generation, even when only local edits are needed. In contrast, the masking mechanism in Mask Diffusion Models (MDMs) n
中文介绍 探索多轮反射掩码机制激发掩码扩散模型的推理能力。相比自回归模型，通过局部编辑实现更高效的修正。
👍 8
Retrieval-augmented generation (RAG) systems must balance retrieval granularity with contextual coherence, a challenge that existing methods address through LLM-guided chunking, single-level context expansion, or hierarchical summarization. These approaches variously depend on costly LLM calls durin
中文介绍 提出SproutRAG，注意力引导的树搜索与渐进嵌入，用于长文档RAG。无需LLM调用即可平衡检索粒度与上下文连贯性。
👍 8
Embodied foundation models are expected to benefit from data scaling like large language models, but face a much tighter data bottleneck. Teleoperated real-robot trajectories remain the dominant pretraining source due to their precise action supervision and embodiment alignment, yet their scalabilit
中文介绍 提出HumanScale，证明第一人称人类视频在具身预训练中可超越实际机器人数据。通过大规模数据克服机器人数据瓶颈。
👍 7
FP4 training promises substantial reductions in memory and computation cost for LLM pretraining, yet current FP4 hardware paths and recipes, including NVIDIA Blackwell/Rubin-class systems and AMD MI350-series GPUs, remain centered on E2M1 data elements. In this study, we identify a fundamental limit
中文介绍 重新审视LLM FP4预训练中的收缩偏差，揭示其几何起源与系统性影响，并提出UFP4配方以改善训练效果。
@AnatoliKopadze · 83.0K 粉丝 · 1.3M 阅 · 584 赞 · 70 转
AI has been in everyone's hands for years. Most people who use it every day still use it the slowest way there is: type a request, wait, fix it, ask again, all by hand. Not because the faster way is
中文介绍 对比Claude、GPT、Mira的Loop实现，指出大多数人仍在手动反复调优，而利用Loop可大幅提升效率。关键在于将交互模式从“提问-修正”转向自动化循环。
@undefinedKi · 3.9K 粉丝 · 1.0M 阅 · 601 赞 · 78 转
Your best ideas are scattered across a dozen places right now. Notes apps. Browser tabs. Old chats with Claude that you closed and will never find again. Every time you sit down to work, you rebuild
中文介绍 用Claude和Obsidian构建AI第二大脑的完整指南，将散落各处的笔记、对话整合为持续进化的知识系统，每次创作不再从零开始。
@eng_khairallah1 · 69.3K 粉丝 · 678.7K 阅 · 506 赞 · 81 转
Most people are using Claude Opus 4.8 to answer questions. Save this :) A small group of people have it running businesses while they sleep. The difference is not the model. You both have access to
中文介绍 分享40个Claude Opus 4.8工作流，用于自动化赚钱，强调差别不在于模型本身而在于如何使用。已有人用它实现“睡着时也能赚钱”。
@sairahul1 · 118.7K 粉丝 · 581.3K 阅 · 516 赞 · 110 转
Learning anything today is easy and confusing at the same time. Easy because AI can explain almost anything in seconds. Confusing because most people just ask random questions, get random answers, and
中文介绍 用Claude加速学习10倍的方法，针对人们常犯的随机提问问题，提倡结构化提问和主动学习，而非被动获取零散答案。
@Oracle_Trade_ai · 39.9K 粉丝 · 197.8K 阅 · 2.8K 赞 · 580 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE官方AI代理在Polymarket上交易，2026年自主AI代理已占预测市场30%以上活动，成为最有效策略之一。
@OracAItrading · 31.8K 粉丝 · 141.6K 阅 · 2.8K 赞 · 576 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE官方AI代理在Polymarket上交易，2026年自主AI代理已占预测市场30%以上活动，成为最有效策略之一。
@Oractrading · 33.9K 粉丝 · 109.2K 阅 · 2.8K 赞 · 585 转
In 2026, autonomous AI agents have become one of the most effective strategies on prediction markets. Over 30% of all activity on Polymarket now comes from algorithmic and AI-powered wallets. We
中文介绍 ORACLE官方AI代理在Polymarket上交易，2026年自主AI代理已占预测市场30%以上活动，成为最有效策略之一。
@mvanhorn · 35.2K 粉丝 · 102.4K 阅 · 510 赞 · 56 转
Earlier this month I wrote WTF Is a Loop? Peter Steinberger vs. Boris Cherny, which did 3.6M views on what a loop even is. This is the sequel, and it answers the next question: which loops do people
中文介绍 Loops系列第二篇，基于前作360万观看，列出15种实际运行的Loops及其命令，并手把手教如何“偷取”这些高效循环。
@PeterDiamandis · 400.3K 粉丝 · 51.7K 阅 · 551 赞 · 75 转
Elon described the near future as a “supersonic tsunami”: a wave moving so fast and so large that by the time you hear it coming, it has already broken over you. The phrase stuck with me. So let me
中文介绍 马斯克将未来5年AI发展比作“超音速海啸”，彼得·戴曼迪斯进一步阐释其含义：变革之快，待你察觉时已被淹没，需立即行动。
@phosphenq · 12.4K 粉丝 · 49.4K 阅 · 506 赞 · 63 转
Last December, Boris Cherny shipped 259 pull requests in a single month. Every one written by Claude. He says he did not open an editor the whole time. His job was not to write code. It was to write
中文介绍 以Boris Cherny用Claude一个月提交259个PR为例，讲解Loop工程核心：不写代码，而是写Prompt，让AI在睡眠时完成编码工作。
OpenAI boardmember Zico Kolter and Gray Swan CEO Matt Fredrikson join swyx to explain why AI security is not just “cybersecurity with AI”
中文介绍 OpenAI董事会成员Zico Kolter与Gray Swan首席执行官Matt Fredrikson共同探讨AI安全，指出其并非简单的“网络安全+AI”，而需深入理解模型行为。
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. For those of you enjoying your summer unaware of Anthropic’s latest feud with the US government, here’s a recap: In April the company said it had built an AI mode
中文介绍 Anthropic与美国政府的最新争端中，本文列出三个值得关注的方面。公司此前于4月做出相关表态。
中文介绍 PP-OCRv6模型上线Hugging Face，支持50种语言OCR，参数量范围1.5M至34.5M。
OpenAI introduces new Daybreak tools, including Codex Security and GPT-5.5-Cyber, to help organizations find, validate, and patch vulnerabilities at scale.
中文介绍 OpenAI推出Daybreak系列工具，包括Codex Security和GPT-5.5-Cyber，帮助组织规模化发现、验证和修补漏洞。
OpenAI introduces Patch the Planet, a Daybreak initiative helping open-source maintainers find, validate, and fix vulnerabilities with AI and expert review.
中文介绍 OpenAI发起Daybreak计划下的Patch the Planet项目，通过AI和专家评审协助开源维护者发现、验证并修复漏洞。
Learn how Jason Liu uses Codex to preserve context, manage complex projects, and help work continue beyond a single prompt.
中文介绍 Jason Liu演示如何利用Codex保持上下文、管理复杂项目，使工作能跨越单次提示持续进行。
中文介绍 TLDR AI简报涵盖编排模型、DeepMind人才外流以及循环工程等话题。
Samsung Electronics deploys ChatGPT Enterprise and Codex to employees worldwide, marking one of OpenAI’s largest enterprise AI rollouts.
中文介绍 三星电子在全球范围内向员工部署ChatGPT Enterprise和Codex，此为OpenAI规模最大的企业AI部署之一。
special offer for subscribers - $250 off AI Engineer tix til Monday
中文介绍 Latent Space为订阅者提供AI Engineer门票优惠，至周一前可享受250美元折扣。
a quiet day lets us promo AIE one last time
中文介绍 AI新闻简报显示当日市场平静，并最后一次推广AI Engineer活动（AIE）。
The Miami-based AI startup Subquadratic came out of stealth mode last month with a huge claim. It announced that it had solved a mathematical bottleneck that had been holding back large language models for almost a decade. The details were thin, and many people were unconvinced. But Subquadratic has
中文介绍 迈阿密AI初创公司Subquadratic宣布解决困扰大语言模型近十年的数学瓶颈，但细节匮乏，未获广泛信服。
With GLM-5.2 passing everyone's vibe check, the open models story finally becomes a real frontier story.
中文介绍 GLM-5.2模型通过普遍检验，开放模型开始成为前沿领域的重要故事；Z.ai预测Open Fable将于12月发布。
**GLM-5.2** emerges as a leading open-weight coding model rivaling **Opus 4.8** and **GPT-5.5** in software engineering tasks, emphasizing the strategic importance of open models for provider competition, on-prem deployment, and fine-tuning rights. Experts like **Patrick Toulme** and **Thomas Wolf**
中文介绍 GLM-5.2成为领先的开源权重编码模型，在软件工程任务中与Opus 4.8和GPT-5.5竞争，凸显了开源模型在竞争、本地部署和微调方面的战略意义。
中文介绍 TLDR AI简报涵盖GPT-5.6将于周二发布、Claude Code的工件功能以及Perplexity的Brain记忆系统。
中文介绍 MosaicLeaks研究探讨研究智能体能否保守秘密，涉及信息泄露风险。
当前市场技术面呈现明显分化。美股方面，QQQ和SPY虽保持多头排列但RSI处于中性区间（52.5/58），而MSFT、META、GOOGL等权重股已转空头，跌幅较大（MSFT近5日-5.99%，RSI 30.9），整体承压。加密资产全面偏空，BTC、ETH、SOL均处于空头排列，RSI 40-47，结合加密恐慌贪婪指数仅20（极度恐慌）及总市值24h微增0.91%但主导率56.2%显示资金集中，弱势明显。中概股集体走弱，BABA、PDD、JD、0700.HK均空头排列且RSI接近或触及超卖（BABA RSI 23），接近52周低点。商品外汇方面，黄金期货（RSI 38.1）与原油（RSI 29.2 超卖）下行，而美元指数（RSI 72.1 超买）与10Y美债收益率（多头排列）走强，宏观信号矛盾。VIX恐慌指数（RSI 48.5）横盘，市场情绪未极端化。整体而言，除美债与美元外，多数资产技术形态偏弱。
现价104.97，RSI14=23.0处于超卖区，价格远低于SMA20(119.59)和SMA50(128.85)，MACD=-6.6384低于信号线-5.1974，空头排列明确，且接近52周低点，空头动能强，暂无反转信号。
现价208.65，RSI14=48.7处于中性区域，MACD=-1.0347略低于信号线-0.2971，柱状线负值微弱，价格在SMA20(211.25)和SMA50(209.81)附近纠缠，无明确方向信号，呈现横盘震荡。
现价101.00，RSI14=72.1超买区，MACD金叉且柱状线正值（0.1033），价格站上所有均线（SMA20=99.68, SMA50=98.97, SMA200=98.70），接近52周高，多头排列强势，显示上行趋势延续。
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
过去走势不代表未来表现，本报告仅供技术指标解读参考，不构成任何投资建议。投资者应结合自身风险承受能力与基本面分析独立决策。
Vice President JD Vance said Iran had agreed to invite U.N. nuclear inspectors into the country, but the Iranian side said it had made “no new commitments.” Inspections had been limited since President Trump tore up an Obama-era deal in 2018.
中文摘要 美国放松对伊朗石油制裁。副总统万斯称伊朗同意邀请联合国核检查员，但伊朗否认作新承诺。自2018年特朗普退出伊核协议后检查受限。
Considered the country’s most powerful leader after the Castro brothers, he was the first director of the Interior Ministry, keeping a close eye on dissent.
中文摘要 古巴监视国家设计师拉米罗·巴尔德斯·梅内德斯去世，享年94岁。他被视为卡斯特罗兄弟后最有权势人物，曾任内政部首任部长，严密监控异见者。
Senator says he wants to see further changes for small businesses and ensure loopholes for wealthier people are actually closed. Follow today’s news live Get our breaking news email, free app or daily news podcast Ben Roberts-Smith invitation to war memorial event ‘appropriate’, says Marles Richard
中文摘要 澳大利亚参议员大卫·波科克批评税收审查荒谬，希望改革小企业税并堵富人漏洞。保琳·汉森解释带薪育儿假立场。
Iran's foreign ministry says it made "no new commitments" on nuclear inspections after talks in Switzerland.
中文摘要 美国放松石油制裁，伊朗否认副总统万斯关于核检查员的说法。伊外交部称在瑞士会谈后未作新承诺。
Gaza’s surf community was nearly wiped out by Israel’s genocidal war. But some surfers continue to chase the waves.
中文摘要 加沙冲浪社区在以色列战争中被摧毁，但部分冲浪者仍追逐海浪寻求慰藉。
A police officer, a civilian and the suspected gunman were killed after a shooting in Montreal.
中文摘要 加拿大蒙特利尔发生枪击事件，一名警察、一名平民和疑似枪手死亡，引发全市避难警报。
After fighting Belgium to a draw for one of Iran’s best World Cup results, the Iranians had a note for their fans and the world.
中文摘要 伊朗队在世界杯上逼平比利时后留便条称“伊朗精神依然鲜活坚定”。此举被视为伊朗最佳世界杯成绩之一。
Keir Starmer, the British prime minister, could be replaced through a full leadership contest. But if the top contender, Andy Burnham, is the only candidate, the process will be much faster.
中文摘要 英国首相斯塔默可能通过领导权竞争被替换。若热门候选人安迪·伯纳姆为唯一候选人，流程将加快。
A police officer and a civilian also died in a gun battle that prompted a citywide shelter-in-place alert.
中文摘要 蒙特利尔枪击案致三人死亡：一名嫌疑人、一名平民和一名警察。枪战引发全市避难警报。
The record company executive also worked with Billy Joel, Pink Floyd, Alicia Keys and Aerosmith.
中文摘要 音乐大亨克莱夫·戴维斯去世，享年94岁。他曾与惠特尼·休斯顿、布鲁斯·斯普林斯汀、比利·乔尔等众多巨星合作。
Committee leaders did not release details but said the legislation would 'hold Big Tech accountable'.
中文摘要 美国众议院委员会达成两党协议，制定社交媒体儿童保护规则。未公布细节，称将追究大型科技公司责任。
An activist in Japan was surrounded by police after he approached the Israeli Embassy in Tokyo on Monday and shouted.
中文摘要 日本一活动人士在东京以色列大使馆外高喊“停止种族灭绝”，被警方制止。
His departure in the coming weeks clears a path for Andy Burnham, a popular Labour Party mayor, to become the country’s seventh prime minister in a decade.
中文摘要 英国首相斯塔默辞职，数周内将由工党市长安迪·伯纳姆接任，成为十年内第七位首相。
The raft of proposed changes includes a new constitution and anti-corruption office, and the ousting of the president.
中文摘要 匈牙利总理发起运动，旨在摆脱欧尔班“黑帮”控制，提议新宪法、反腐败办公室并罢免总统。
Plan to put funding under direct control has sparked widespread opposition, with critics fearing political interference.
中文摘要 捷克公共媒体员工罢工，反对政府将媒体资金直接控制的计划，担忧政治干预独立性。
Gold steadied after the US and Iran flagged early progress in negotiations to end the war that’s upended global markets and stoked inflation.
中文摘要 黄金价格企稳，此前美国和伊朗表示在和谈中取得初步进展，这场战争曾扰乱全球市场并推高通胀。
Iran's foreign ministry says it made "no new commitments" on nuclear inspections after talks in Switzerland.
中文摘要 美国放松对伊朗石油制裁，但伊朗外交部称在瑞士会谈中未就核检查作出新承诺。
Hydration tracking gadgets are flooding the market but is it too much information?
中文摘要 补水追踪设备涌入市场，可提供「出汗分数」等指标，但信息过量问题引发关注。
The UK will have secure power supplies this winter despite the disruption to energy markets from the Iran war, according to the National Energy System Operator.
中文摘要 英国国家能源系统运营商称，尽管伊朗战争扰乱能源市场，今冬英国电力供应仍将有保障。
USA Rare Earth Inc. pushed back against allegations by MP Materials Corp., calling a lawsuit filed by its rival an attempt to hinder the company’s growth.
中文摘要 USA Rare Earth公司反击MP Materials的诉讼，称对方试图阻碍其发展，两家公司竞争加剧。
A Tesla driver on Friday said he was using ‘self-driving’ technology after his vehicle sped into a Texas home.
中文摘要 美国联邦调查一起特斯拉致命车祸，司机称事发时正使用「自动驾驶」技术，车辆冲入得州一处住宅。
Asian stocks were set for a positive open as oil largely held Monday’s losses, with investors continuing to watch developments in US-Iran peace talks.
中文摘要 亚洲股市有望开盘上涨，油价基本持稳于周一的跌幅水平，投资者关注美伊和谈进展。
Oil edged higher after falling almost 3% in the previous session following early progress in peace talks over the Iran war, which included a US waiver allowing some sales from the Islamic Republic.
中文摘要 油价小幅上涨，前一交易日因美伊和谈取得初步进展（包括美国豁免部分伊朗石油销售）而下跌近3%。
Avis Budget Group Inc. is set to receive $650 million in cash as part of a settlement agreement with Pentwater Capital Management to resolve a lawsuit regarding short-swing profits, according to a filing.
中文摘要 根据和解协议，Avis Budget Group将从Pentwater Capital Management获得6.5亿美元现金，以解决有关短线交易利润的诉讼。
Bending Spoons SpA, which acquires struggling software businesses, and some of its backers are seeking to raise as much as $1.62 billion in an initial public offering.
中文摘要 Vimeo母公司Bending Spoons及其部分投资者寻求通过首次公开募股（IPO）筹资高达16.2亿美元。
Edgewell Personal Care Co., which makes Schick razors, has rejected an unsolicited takeover offer from private equity firm Yellow Wood Partners, according to people familiar with the matter.
中文摘要 据知情人士称，Schick剃须刀制造商Edgewell Personal Care已拒绝私募股权公司Yellow Wood Partners的主动收购要约。
Lindsay Shumlas, CEO of Cotopaxi, discusses the company's recent expansion from backpacks into the luggage market with the launch of their new Coraza luggage line. The product, introduced about two months ago, emphasizes sustainability and durability through a repair-led model, contrasting with the
中文摘要 Cotopaxi公司CEO Lindsay Shumlas表示，推出Coraza行李箱系列是品牌的自然延伸，注重可持续性和耐用性，采用以修复为主导的模式。
22 回复 · 程序员 节点
6 回复 · 程序员 节点
7 回复 · 程序员 节点
6 回复 · Apple 节点
16 回复 · 程序员 节点
10 回复 · Apple 节点
26 回复 · Apple 节点
11 回复 · Rust 节点
23 回复 · Apple 节点
54 回复 · Apple 节点
从虽然是女生 我也整点??继续讨论： 这次有衣服了!! 更新一句: 这些就是更新的内容了 下面为原内容 再说:本人实在讨厌黑丝等暗色袜子(除了肉色好一点) 以及破洞等表面不纯的袜子(暖暖穿的除外) （点击了解更多详细信息） 就这样啦 只是觉得真的好麻烦 拼尽全力尝试让图片更好看一点点 但总感觉效果不好，我似乎毫无审美细胞 这段时间全靠生理本能驱动了 放弃思考 今天(6月21号至今此时)自我评价为时间最没去到该去的地方的一天 25 个帖子 - 24 位参与者 阅读完整话题
从CC接入gpt模型现在不自动压缩了继续讨论： 有人问我说 这么个问题 说不压缩了 还给了个issue 故意恶心你的 github.com/anthropics/claude-code Auto-compact stopped working for third-party API providers since v2.1.161 06:48AM - 05 Jun 26 UTC ### Preflight Checklist - [x] I have searched [existing issues](https://github.…com/anthropics/claude-code/is
一天在忙， 又老一岁了，时间过得真快 老登了， 希望家人身体健康， 佬友们天天开心 41 个帖子 - 41 位参与者 阅读完整话题
很早之前因为看到一个抽奖手贱，然后评论了一下被举报了，接着惨遭禁言，一直不知道怎么解除，然后就只能看帖子却不能回复，不能参与到各位佬的讨论当中难受极了，今天突然想到可以找管理员问问，然后写了百字小作文，诉说了我不能交流的难受之情，没几分钟就解禁了，感谢管理员大大。还有告诫自己以后没事少碰抽奖这个东西 58 个帖子 - 50 位参与者 阅读完整话题
这两天看了太多同学们的简历，50份有没有…看得出大家还是很有要性的，我当初选择Linuxdo就没选错。但是还是很多同学们的简历中就是有一些demo项目，也不知道怎么突破，怎么办 所以我拿出我压箱底的东西，给大家看看。校招的同学们如果没有Agent项目，尽量往这三点上靠，简历的质量能有一个较大的提升 逆向一个没有公开文档的 API 找一个没有公开 API 的网站，逆向分析它真实的工作方式：认证机制、请求结构、限流策略、分页方式、反爬机制等等。然后基于你解锁出来的能力，做一个真正的产品。 这里有两个难点：一是破解系统，二是判断解锁这些能力之后，应该做什么产品。抓一份 CSV 不算产品。你能拿到的数
修好了，睡觉 由于资源紧张，重新限制为三级用户使用 清了6784个邮箱用户 本月已免费提供 967,008,431,220 tokens 所谓有求而不得，人心欲壑，可填沧海 512 个帖子 - 460 位参与者 阅读完整话题
我是一个刚入站不久的小伙伴。应该大部分兄弟都和我有一样的感觉，这里热度最高且最吸引人的就是各种各样的公益站了。最近虽然在期末考试，但主播不到考前两天没有学的欲望哈哈，于是测试了站里的不少公益站。基本感受可以从稳定性和模型能力两方面来看。 1：某地方的GPT 5.5 100分 出题人 （疑似看不惯Claude的做题风格给2扣了两分） 2：某地方的Claude opus 4.8 98分 （我给100，裁判扣分理由可以理解为解题过程不够详细，就像高中数学老师总是叫你把做题步骤写详细，但我就喜欢不写那些每个这种类型题都要写的简单且字多的步骤。。。也许这套题目根本没让Claude大人尽兴？？） 3：网页
灵感来自： 关于 Typora 最新激活脚本-优化版【2026-05-23有效】 资源荟萃 Windows 用户看这里 参考这篇文章，昨天在 Windows 上对最新版 Typora（v1.13.6.0）进行了激活。 但是我遇到了几个问题： 1、首次（后台没有 Typora 进程）打开 Typora 或使用 Typora 打开 md 文件，会自动弹出开发者工具。 2、我把前台窗口关闭后，再次点击打开其他 md 文件，无法打开（无法弹窗），任务管理器查看有 Typora 后台进… 起因是我在两台电脑上都需要激活，然后我把脚本给我的两台电脑上的agent让它们操作，但是操作的时候会有各种各样的问题
根据 DeepSWE 数据，GLM5.2 比 opus4.8 low 强，弱于 opus4.8 medium 另外有几个冷知识 fable 5 low 比 opus4.8 max 更强，而且费用也要便宜很多 gpt 5.5 medium 比 5.4 xhigh 更强，费用更便宜 gpt 5.5 high 费用大概是 5.5 medium 的 1.85 倍 性价比最高的方案是 gpt 5.5 medium 比 opus4.8 high/medium 和 glm5.2 max 更强，更便宜 结论： 37 个帖子 - 31 位参与者 阅读完整话题
先说一下我的情况，两个眼500度近视以内，两眼有100度+的度数差。散光50度左右。 之前的眼镜多数是增加近视度数补偿散光了，导致每天下午3点之后，就会觉得眼压高，眼珠子胀疼流泪。 后来我发现，把系统颜色改为深色模式，眼压高的症状会缓解很多，然后就把系统、各种应用都改了深色模式，手机同时考虑到省电，也都改了深色模式。 我去年配了一个眼镜之后，镜片把散光加上了，就出现了一个新的问题，不看电脑的时候，看别的东西变得模糊了，感觉眼睛对焦很慢，而且模糊。我以为是散光更严重了，而且不知道是什么引起的。 然后就一直看不清，用了滴眼液吃了叶黄素，也没啥太大作用。 直到上周，我才知道，深色模式会让眼睛可能因瞳