TypeScript · ★ 35,638 · 🍴 2,863 · 📈 4,721 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式知识图谱的工具，支持探索、搜索和问答。区别于仅展示用途的静态图谱，它更注重教学价值，帮助开发者快速理解陌生代码结构。兼容 Claude Code、Codex、Cursor、Copilot、Gemini CLI 等主流 AI 编程工具，适合需要快速上手大型项目或进行代码审查的工程师使用。
JavaScript · ★ 194,281 · 🍴 29,959 · 📈 1,912 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 编程代理的性能优化框架，提供技能管理、本能行为、记忆持久化、安全策略和研究优先开发等能力。旨在提升 Claude Code、Codex、Opencode、Cursor 等代理的整体执行效率，适合需要构建或优化复杂 AI 代理工作流的开发者使用。
Python · ★ 20,646 · 🍴 3,443 · 📈 2,169 stars today
Learn it. Build it. Ship it for others.
中文介绍 从零开始的 AI 工程实战课程仓库，涵盖学习、构建到上线发布的完整流程。面向希望系统掌握 AI 工程技能的开发者，强调动手实践与真实项目交付，适合自学入门或提升 AI 应用开发能力的工程师参考使用。
Python · ★ 16,628 · 🍴 1,952 · 📈 1,698 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，专为知识工作者在 Claude Cowork 环境中使用而设计。插件覆盖典型知识处理场景，帮助用户扩展 Claude 的工作流能力。研究人员、分析师、内容创作者等可借助这些插件在 Claude 中完成更复杂的知识管理与处理任务。
Python · ★ 10,068 · 🍴 1,182 · 📈 871 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 代理提供 754 项结构化网络安全技能的标准化库，映射至 MITRE ATT&CK、NIST CSF 2.0、MITRE ATLAS、D3FEND 和 NIST AI RMF 五大权威框架，遵循 agentskills.io 标准。兼容 Claude Code、GitHub Copilot 等主流 AI 工具，适合安全研究人员和构建网络安全 AI 代理的开发者使用。
★ 4,984 · 🍴 401 · 📈 547 stars today
A skill file for removing AI tells from prose
中文介绍 专为消除 AI 写作痕迹而设计的技能配置文件（skill file），帮助 AI 生成的文本更接近自然人类写作风格。通过特定规则约束 AI 输出，去除常见的模板化表达和机械感措辞，适合需要对 AI 生成内容进行润色或直接输出高质量文章的写作者和内容创作者。
Shell · ★ 21,493 · 🍴 1,731 · 📈 1,440 stars today
Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop
中文介绍 通过自定义技能配置赋予 AI 更好的审美品味，抑制其生成平淡、通用、千篇一律的内容。作为 AI 写作或创意辅助工具的补充配置，适合对输出质量有较高要求、希望 AI 产出更具个性和风格的创作者与开发者使用。
HTML · ★ 167,256 · 🍴 3,078 · 📈 1,127 stars today
DigitalPlat FreeDomain: Free Domain For Everyone
中文介绍 DigitalPlat 推出的免费域名申请平台，面向所有人开放，无需付费即可获取域名。适合个人开发者、学生或预算有限的小型项目用于搭建网站、测试服务或展示作品，降低互联网入门门槛。
C# · ★ 52,363 · 🍴 4,870 · 📈 91 stars today
The Free Software Media System - Server Backend & API
中文介绍 完全免费开源的个人媒体服务器，提供服务端后端与完整 API，是 Plex、Emby 等商业媒体服务器的开源替代方案。支持影视、音乐、图片等多种媒体类型的管理与流媒体播放，无追踪、无订阅费用，适合希望自托管家庭媒体库的用户部署使用。
JavaScript · ★ 5,227 · 🍴 267 · 📈 738 stars today
Curated list of the best free apps for PC and mobile
中文介绍 精心整理的 PC 端与移动端优质免费应用清单，涵盖多个使用场景分类。为用户提供一站式免费软件参考，避免为常见需求付费购买商业软件，适合寻找生产力工具、创意软件或日常应用替代品的普通用户和开发者查阅。
TypeScript · ★ 46,828 · 🍴 6,649 · 📈 237 stars today
The open alternative to Salesforce, designed for AI.
中文介绍 专为 AI 时代设计的开源 CRM 系统，定位为 Salesforce 的开源替代品。提供客户关系管理的核心功能，并在架构层面为 AI 集成预留空间。适合希望自主掌控数据、降低 SaaS 成本，同时拥抱 AI 驱动销售流程的中小企业和技术团队部署使用。
TypeScript · ★ 12,073 · 🍴 1,625 · 📈 128 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 开源的股票市场数据平台，提供实时价格追踪、个性化价格预警和公司深度信息查询功能，是昂贵商业行情平台的免费替代方案。完全开放、永久免费，面向个人投资者、金融数据爱好者及希望在此基础上进行二次开发的开发者。
TypeScript · ★ 78,612 · 🍴 6,763 · 📈 319 stars today
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
中文介绍 为 AI 代理提供跨会话持久化记忆能力的工具，能自动捕获代理在每次会话中的操作内容，使用 AI 进行压缩摘要，并在未来会话中智能注入相关上下文。兼容 Claude 及多种主流 AI 代理框架，解决代理每次启动时"失忆"的痛点，适合构建有状态长期任务代理的开发者。
Python · ★ 3,467 · 🍴 247 · 📈 207 stars today
A contact solver for physics-based simulations involving 👚 shells, 🪵 solids and 🪢 rods.
中文介绍 面向物理仿真场景的高性能接触求解器，支持薄壳（衣物）、固体和杆件三类物体之间的接触碰撞计算。采用 PPF（Point-Point-Face）方法提升求解精度与稳定性，适用于布料模拟、软体动力学等计算机图形学与工程仿真领域的研究人员和开发者。
👍 15
Reinforcement learning with verifiable rewards (RLVR) has driven breakthroughs in domains such as math, tool-use, and software engineering, yet its extension to computer-use agents (CUAs) has been bottlenecked by the scarcity of scalable training data with deterministic rewards. Constructing such da
👍 6
Transformer-based large language models are increasingly used for long-horizon tasks; however, their attention mechanism scales poorly with context length. To handle this, we study a sleep-like consolidation mechanism in which a model periodically converts recent context into persistent fast weights
👍 12
While AI agents demonstrate remarkable capabilities in reasoning and tool use, they remain fundamentally reactive: they compute responses only after explicit user prompts. This paradigm ignores a critical opportunity: the idle time between interactions is largely wasted, leaving agents unable to pre
👍 7
In this paper, we introduce InstructSAM, a unified and streamlined framework designed for multi-instance segmentation under arbitrary instructions. We formulates instruction-driven instance segmentation as a set-structured query prediction problem and propose an explicit reasoning-to-instance query
👍 5
Current video-to-4D methods struggle with complex topology changes, transparent materials, thin structures, and inner surfaces. We present Helix4D, a dynamic mesh generation framework by inheriting the expressive representation of Trellis2, adapting it from image-to-3D to video-conditioned 4D genera
👍 34
Sparse-view 3D reconstruction is increasingly addressed with feed-forward splatting networks that predict explicit primitives directly from images. Yet most existing methods remain centered on Gaussian primitives and expose surfaces only indirectly: extracting a usable mesh for downstream simulation
👍 12
Large language model agents are increasingly envisioned as always-on personal assistants with access to anything relevant in the user's digital world. Yet current systems operate over only narrow slices of that world, limiting context-sensitive reasoning and effective assistance. Existing benchmarks
👍 1
Sparse encoders offer high-precision retrieval by representing term importance within a vocabulary space, yet their English-centric structures pose a critical impediment to language transfer for non-English languages. To overcome this structural limitation, we propose SemBridge, a novel embedding in
👍 31
Multimodal modeling represents a vital step from modality-agnostic reasoning toward world modeling. While early approaches predominantly rely on late-fusion that assembles encoders and frozen language backbones with output heads, recent efforts have shifted the paradigm toward native multimodal mode
👍 1
Reward hacking arises when a model improves a proxy reward by exploiting shortcuts rather than solving the intended task. We study this failure mode through the geometry of reinforcement learning updates in language models and argue that hacking emerges when optimization drifts away from a stable lo
👍 1
Mobile GUI agents powered by large language models have progressed rapidly, creating urgent needs for realistic and comprehensive evaluation. Existing benchmarks prioritize reproducibility but are often limited to open-source apps or file-operation tasks for the difficulty of constructing rewards on
👍 0
Long-form image captioning exposes a reward granularity problem in RL: captions are judged as whole sequences, while the important errors occur at the level of individual visual claims. A good dense caption should be both faithful and informative, avoiding hallucination without omitting salient deta
👍 1
CLI agents are the closest thing language models have to an embodied setting: the model emits commands, the terminal executes them, and the returned stream -- stdout, errors, files, logs, and traces -- records the consequences. We argue that this stream is a supervision signal, but standard agent RL
👍 3
Large Language Model (LLM) agents are increasingly improved through interaction, yet most self-evolution methods adapt either the policy or the learning environment in isolation. We identify this structural gap as Agent-Environment Misalignment: the agent's capability frontier changes during trainin
👍 2
Recently, Reinforcement Learning with Verifiable Rewards (RLVR) and Test-Time Scaling (TTS) have advanced LLM code generation through executable verification. Yet Ground-Truth Unit Tests (GT UTs) remain a bottleneck: SOTA RLVR methods require them for costly training, while existing TTS methods lose
👍 1
Online 3D reconstruction requires estimating camera pose and scene geometry under strict causal and bounded-memory constraints. Existing methods often suffer from drift, jitter, or collapse on long sequences. We trace these failures to a fundamental mismatch. Streaming geometry is inherently tempora
👍 0
Large Reasoning Models (LRMs) exhibit backtracking and self-verification mechanisms that enable them to revise intermediate steps and reach correct solutions, yielding strong performance on complex logical benchmarks. We hypothesize that such behaviors are beneficial only when the model has sufficie
👍 11
Large language model (LLM) agents accumulate rich episodic trajectories while solving real-world tasks, but it remains unclear whether such experience can be distilled into reusable procedural skills. We introduce SkillEvolBench, a diagnostic benchmark for evaluating this step from experience reuse
👍 28
Deep research agents extend the role of search engines from retrieving keyword-matched pages to synthesizing knowledge, fundamentally changing how humans interact with information. However, frontier systems remain proprietary, while existing open agents often generalize poorly across different task
👍 58
Autonomous agents are moving from tools into a layer of social infrastructure: they browse, purchase, deploy software, manage systems, and increasingly interact with one another. As these systems scale, the bottleneck shifts away from raw model capability toward coordination. Agents need to form rel
👍 0
Current motion-controlled image-to-video generation models rigidly follow user-provided trajectories that are often sparse, imprecise, and causally incomplete. Such reliance often yields unnatural or implausible outcomes, especially by missing secondary causal consequences. To address this, we intro
👍 1
Temporal credit assignment in reinforcement learning has long been a central challenge. Inspired by the multi-timescale encoding of the dopamine system in neurobiology, recent research has sought to introduce multiple discount factors into Actor-Critic architectures, such as Proximal Policy Optimiza
👍 28
Efficient attention algorithms are critical to mitigate the quadratic cost of attention in long-context workloads. Prior work utilises block-scaled quantisation techniques on Blackwell GPUs to move attention computation to 4-bit precision to accelerate inference. However, these techniques result in
👍 1
As large language model (LLM)-based agents increasingly participate in online discourse, red-teaming their capacity to support political influence campaigns is critical for information integrity. In pursuit of this goal, we focus on locally deployed open-source LLMs, as opposed to frontier API-only
👍 1
As interactive LLM-based applications are created and refined, model developers need to evaluate the quality of generated text along many possible axes. For simpler systems, human evaluation may be practical, but in complicated systems like conversational chatbots, the amount of generated text can o
👍 1
Transportation safety analysis requires integrating crash records, roadway attributes, and geospatial data through GIS-based workflows, but access remains uneven across agencies and community stakeholders. Technical prerequisites create a gap between analytical tools central to safety planning and t
👍 1
Communication enables coordination in multi-agent reinforcement learning (MARL), but many real-world applications, e.g., search-and-rescue with drone swarms, operate under severe bandwidth constraints. Many communication architectures still expose a coupled bottleneck in which a shared latent repres
👍 29
Training large multimodal models (LMMs) via reinforcement learning (RL) to natively invoke video-processing tools (e.g., cropping) has become a promising route to long-video understanding. However, existing native-RL methods dispatch tool calls sequentially (i.e., one per turn): a single wrong crop
👍 8
Memory is a fundamental component for enabling long-context LLM agents, supporting persistent state across interactions through a continuous serve-and-update lifecycle. Despite substantial prior work, existing systems suffer from significant maintenance overhead due to two key limitations: coarse-gr
👍 4
Unified multimodal understanding and generation models enable richer human-AI interaction. Yet jointly customizing a character's persona, dialogue style, and visual identity while maintaining output consistency across modalities remains largely unexplored. To mitigate this gap, we introduce a new ta
@sairahul1 · 106.0K 粉丝 · 2.0M 阅 · 1.1K 赞 · 136 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 AI Agent 协作系统，旨在实现「软件工厂」式自动化开发——让功能在你睡觉时持续交付。作者区分了「用 AI 辅助打字」与「真正自动化工程流水线」的本质差异，并梳理了该多 Agent 架构的设计思路与落地方法。
@sairahul1 · 106.0K 粉丝 · 203.2K 阅 · 500 赞 · 82 转
Everyone is talking about AI agents in 2026. Most people have no idea how they actually work. This changes today. I spent weeks distilling everything: courses, books, real builds, production failures.
中文介绍 作者耗数周整理了一套 AI Agent 完整学习课程，涵盖课程、书籍、实际项目构建与生产环境踩坑案例，面向真正想搞懂 Agent 底层运作机制的人。内容定位为从零到生产级的系统性梳理，而非泛泛科普。
@cyrilXBT · 179.9K 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 教程分享：如何用 Claude 搭建一个晨间自动研究 Agent，每天自动扫描互联网内容并在 5 分钟内生成简报，替代手动刷 Twitter 和邮件的低效信息摄取方式。核心价值在于将信息筛选流程完全自动化，解放早晨注意力。
@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 Cursor 员工坦言：入职面试前从未用过 Cursor，此前在 Meta 一直重度使用 Claude Code，甚至自掏 200 美元/月订阅。帖子记录了他从 Claude Code 切换到 Cursor 的真实体验与使用心得，属于难得的内部视角对比。
@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 工程实战项目路线图：从零手写 Tokenizer 开始，依次实现 Embedding、位置编码、Attention、Transformer Block、训练目标、解码策略、KV Cache 到长上下文处理。强调「动手造轮子」而非停留在阅读理解层面。
@hansolar21 · 30.1K 粉丝 · 50.8K 阅 · 500 赞 · 82 转
tldr: Korea is already being repriced as an AI memory market. But I think the next leg could be Physical AI: robots, actuators, reducers, and the boring precision-machining companies sitting
中文介绍 投资视角分析：韩国已作为 AI 存储市场被重新定价，但作者认为下一波机会在 Physical AI——机器人、执行器、减速器及精密机加工零部件供应商。重点关注那些低调却处于机器人产业链关键节点的韩国零部件企业。
@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 回顾 2024 年 8 月 Andrej Karpathy 发布的一个 40 行 Shell 脚本：并非论文或模型，却意外演变为重要基础设施。帖子分析这段代码为何影响深远，探讨简洁工具在 AI 工程实践中的传播与演化路径。
@0xileri · 7.3K 粉丝 · 12.2K 阅 · 533 赞 · 63 转
I’ve been getting a lot of DMs since I started posting AI videos, so I figured I’d just write it all out. Fair warning: I’m still learning too. This is just what’s been working for me. tools
中文介绍 新手向 AI 视频制作工具链分享：作者因收到大量私信而整理出自己的实际工作流，坦诚「仍在学习中」，列出目前有效的工具组合与制作步骤。适合刚入门 AI 生成视频、想了解真实操作流程的创作者参考。
@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，主张 AI 替代人类劳动已从预测变为现实，正在引发「现代史上最大经济转型」。整体以煽动性叙事为主，缺乏具体数据或案例支撑，属于 AI 冲击就业话题的常见论调，信息密度偏低。
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
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—
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
OpenAI partners with Grupo Folha and Grupo UOL to bring trusted Brazilian journalism to ChatGPT, expanding access to news with attribution and transparency.
a quiet day lets us tie together a few quotes as all model labs become agent labs
During Tuesday’s Google I/O keynote, Demis Hassabis, the CEO of Google DeepMind, proclaimed that we are currently “standing in the foothills of the singularity.” It was a striking statement—the singularity is the theoretical future moment when AI rapidly exceeds human intelligence and dramatically t
a quiet day lets us feature fundraises!
How Virgin Atlantic used Codex to ship its revamped mobile app on a fixed holiday travel deadline, reaching near-total unit test coverage and zero P1 defects.
OpenAI is named a leader in the 2026 Gartner Magic Quadrant for Enterprise AI Coding Agents, with Codex recognized for innovation and enterprise-scale deployment.
本期美股整体技术面延续强势格局，但超买信号已在多处触发。SPY 报 750.59 美元，RSI14 71.3，距 52 周高点仅 -0.2%，处于超买区间但趋势标签仍为「bullish」，SMA20/50/200 多头排列完整。QQQ 以 RSI14 75.3 领跑全场，5 日涨幅 3.46%，MACD 死叉雏形初现。AAPL RSI14 77.4 为本期最高，而 MSFT 与 NVDA 的 RSI 回落至 52-53 的中性区间，二者 MACD 均现死叉，短期动量明显弱于大盘。META 价格低于三条均线，MACD 维持负值，技术结构相对偏弱。加密市场方面，恐慌贪婪指数录得 34（恐慌档），加密总市值 2.62 万亿美元，24 小时下滑 1.5%，BTC 主导率 58%。BTC、ETH、SOL 三者均处于 SMA 空头排列，MACD 负值扩张，RSI14 分别落在 42.7、35.7、41.9，整体技术面偏弱。中概股中，腾讯控股（0700.HK）RSI14 33.9 接近超卖临界，价格距 52 周低点仅 +1.62%，空头排列形态突出；BABA、PDD 同样位于均线下方。商品与外汇方面，WTI 原油 5 日跌幅高达 -13.86%，MACD 死叉确认，技术信号转弱最为鲜明；黄金期货价格低于 SMA20 和 SMA50，RSI14 39.2 偏弱。美元指数 DXY 报 99.14，RSI14 55.4，多头排列，但距 52 周高点仅 -1.49%，上方压力不远。10 年期美债收益率 4.49%，多头排列，RSI53.6，整体宏观利率压力尚存。VIX 报 17.01，当日上升 2.53%，但仍低于 SMA50（20.44），短期市场情绪波动轻微抬升。
QQQ 当前报价 730.28 美元，5 日涨幅达 3.46%，距 52 周高点仅 -0.12%，处于多头排列（SMA20 698.21 > SMA50 644.84 > SMA200 615.47）。RSI14 高达 75.3，深度超买区间。MACD 报 20.898 与信号线 21.601 的微小死叉值得关注，但从斜率与价格位置看，动量仍偏强。超买读数意味着短期回调风险上升。
META 报价 612.34 美元，5 日涨幅仅 0.18%，距 52 周高点已回落 -23.1%。SMA20 615.79、SMA50 617.79、SMA200 668.68 构成空头排列，价格低于三条均线。MACD -6.561 低于信号线 -6.169，绿柱延续。RSI14 为 46.4，处于中性偏弱区。整体技术结构呈空头主导态势。
BTC 报价 75,808 美元，1 日跌 -1.9%，5 日跌 -2.23%，距 52 周高点下滑 -39.93%。价格低于 SMA20（78,555）、SMA50（77,076）、SMA200（80,268），呈标准空头排列。MACD -322.34 大幅偏离信号线 121.74，绿柱深度扩张。RSI14 42.7 接近偏弱区间。结合加密恐慌贪婪指数 34（恐慌档）及加密总市值 24h -1.5%，技术面下行压力明显。
WTI 原油报 93.6 美元，1 日重挫 -3.11%，5 日大跌 -13.86%，距 52 周高点回落 -21.66%。价格明显跌破 SMA20（100.85）和 SMA50（98.28）。MACD 0.488 已低于信号线 1.550，触发死叉。RSI14 43.7 位于中性偏弱区间。短期动量加速下行，技术结构转弱信号显著。
AAPL 报价 308.33 美元，5 日涨 3.52%，多头排列完整（SMA20 291.38 / SMA50 271.72 / SMA200 261.99）。RSI14 77.4 为本批标的最高值，严重超买区间。MACD 10.09 高于信号线 9.21，红柱仍正值，但 1 日微跌 -0.16% 且距 52 周高点仅 -1.12%，上行空间受压。超买与强趋势共存，方向信号相互抵消，整体呈中性观察态势。
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
本报告仅依据公开行情数据计算的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推荐。过去走势不代表未来表现，市场存在不确定性，读者需自行承担相应决策风险。
Australia’s former defence secretary, Greg Moriaty, has begun his term as Australia’s ambassador to the US. Follow today’s news live Get our breaking news email, free app or daily news podcast Syrian children should be treated ‘sensitively and gently’, Ryan says Independent MP Monique Ryan says the
中文摘要 澳大利亚前国防部长格雷格·莫里亚蒂正式就任澳大利亚驻美大使。澳大利亚国家反腐委员会（NACC）即将卸任的负责人就机器人债务调查延误一事公开致歉。澳大利亚政坛多项议题持续受到关注，包括求职者福利、资本利得税及预算案等。
The director of Britain’s electronic surveillance agency warns that Russia is only getting more brazen as battlefield losses in Ukraine mount.
中文摘要 英国政府通信总部（GCHQ）负责人发出警告，称随着俄罗斯在乌克兰战场损失不断增加，其在全球范围内的情报与网络行动愈发肆无忌惮，威胁持续扩大，敦促各方高度警惕俄方的进一步挑衅行为。
Iran has warned that it would retaliate after American military forces attacked sites near the Strait of Hormuz.
中文摘要 美国官员表示，美军在伊朗恢复空袭前已掌握伊朗威胁情报。伊朗方面警告将对美军打击霍尔木兹海峡附近目标一事采取报复行动。此次袭击发生于美伊外交谈判持续推进的背景之下，局势高度紧张。
Al Jazeera’s @avawarrinerr explores the rise of the Cockroach Janta Party.
中文摘要 半岛电视台报道了印度新兴网络政治运动"蟑螂人民党"的崛起现象，分析其是否对莫迪领导的印度人民党政府构成政治威胁。该党借助社交媒体快速传播，引发广泛关注。
Several injured people have been transported to hospitals to be treated for chemical burns.
中文摘要 美国华盛顿州发生化学储罐破裂事故，造成多人死亡，另有数名伤者因化学灼伤被紧急送往医院救治。事故具体原因及涉及化学物质种类尚在调查之中。
Authorities in Britain and France warned that “unprecedented” temperatures — at a time when few expected them — could persist. Climate change has made heat waves more frequent and more intense.
中文摘要 英国、西班牙和法国正经历一场打破纪录的春季热浪，英法两国当局警告气温"前所未有"且可能持续。气候科学家指出，气候变化导致热浪更加频繁和强烈，此次极端高温发生时间远早于正常季节。
Dong Guangping, 68, fled China at least three other times but was always sent back. His friends hope things might be different this time.
中文摘要 68岁中国异见人士董广平乘坐橡皮艇出逃，成功抵达韩国。他此前曾至少三次试图逃离中国，但均遭遣返。其友人希望此次韩国方面的处理结果有所不同，不会将其送回中国。
The 36-year-old, named in the French news media as David G., is among more than 70 employees at schools in the capital who have been recently suspended or fired over allegations of sexual abuse and other misconduct.
中文摘要 法国巴黎一名36岁学校工作人员（法媒称其为"大卫·G"）因涉嫌儿童性侵被提起诉讼。此案属于巴黎校园性侵系列调查的一部分，目前已有逾70名学校员工因性侵及不当行为指控遭停职或解雇。
New videos show widespread destruction in Maarakeh following deadly Israeli strikes across southern Lebanon.
中文摘要 最新视频显示，以色列对黎巴嫩南部多地发动致命空袭后，马阿拉凯等地满目疮痍，建筑大面积受损。此次袭击造成人员伤亡，黎巴嫩南部局势持续紧张。
The country has always stood apart from the continent. Then President Trump started threatening Greenland.
中文摘要 冰岛历来在欧洲政治一体化进程中保持距离，但特朗普总统对格陵兰岛发出领土威胁后，冰岛对深化欧洲合作的态度明显转暖，其与欧陆关系或迎来转折点。
The minibus, carrying nine people, drove through closed crossing barriers during the morning rush hour.
中文摘要 比利时发生一起严重交通事故，一辆载有9名乘客的小型客车在早高峰期间穿越已关闭的铁路道口护栏，与火车相撞，造成4人死亡，多人受伤。
Climber carries kite with Palestinian children’s dreams to Everest summit
中文摘要 一名登山者将载有巴勒斯坦儿童梦想寄语的风筝带至珠穆朗玛峰顶峰，以此表达对加沙儿童的声援与关怀，这一举动引发国际社会广泛关注。
The warnings came after U.S. forces struck military sites in Iran. Secretary of State Marco Rubio said on Tuesday that negotiations to end the war were continuing.
中文摘要 美军打击伊朗境内军事目标后，伊朗威胁将发动新一轮反击。美国国务卿马科·鲁比奥周二表示，结束冲突的外交谈判仍在继续。此次紧张局势发生于双方推动外交解决方案的微妙时期。
Extreme early summer event across western Europe also brings highest temperatures for month in UK and Ireland Seven people have died in France in an extreme early summer heat event affecting a swathe of western Europe, with record high temperatures for May recorded for a second day in several countr
中文摘要 席卷西欧的极端初夏热浪已造成法国7人死亡。英国、爱尔兰及多国均录得5月份历史最高气温，创纪录高温连续第二天持续。气象专家警告此次热浪属罕见异常天气事件。
This live blog is now closed. For the latest, read our full report: Iran remains in peace talks despite first US strikes since ceasefire Iranian supreme leader Mojtaba Khamenei said on his Telegram channel that Gulf powers will no longer be a shield for US bases and the US will no longer have a safe
中文摘要 伊朗外交部指责美国在停火后发动夜间空袭，称其违反停火协议。伊朗最高领袖莫吉塔巴·哈梅内伊在其Telegram频道上警告，海湾国家将不再是美国基地的"盾牌"。伊朗表示尽管遭到袭击，仍将继续参与和平谈判。
President Donald Trump backed the Commodity Futures Trading Commission as the sole regulator of prediction markets, siding with the agency amid a growing legal fight with states over the multibillion-dollar industry.
中文摘要 特朗普支持商品期货交易委员会（CFTC）成为预测市场的唯一监管机构，在联邦与各州围绕这一数十亿美元规模行业的监管权争议中，总统明确站队CFTC一方。目前多个州正就预测市场监管权与联邦机构展开法律较量。
FTSE Russell adopted a rule change that will speed the addition of newly listed large-cap companies to its main indexes, weeks ahead of SpaceX’s expected record-breaking initial public offering.
中文摘要 富时罗素宣布修改规则，将加快把新上市大市值公司纳入旗下主要指数的速度。此举距SpaceX预期创纪录的IPO仅剩数周，新规将有助于SpaceX上市后更快获得指数纳入资格。
Asian stocks were set to edge higher Wednesday in a sign traders were clinging to expectations the US and Iran will sign a peace deal even as military strikes in the Persian Gulf persisted.
中文摘要 周三亚洲股市小幅走高，市场对美伊签署和平协议仍抱有期待。尽管波斯湾地区军事冲突持续，投资者整体情绪偏乐观。此前美伊谈判进展及地缘局势走向成为市场主要关注焦点。
Campaigners welcome the move but say passengers are mostly worried about train fares and delays.
中文摘要 英国政府计划为数百列火车升级WiFi网络。倡导团体对此表示欢迎，但指出乘客目前最关心的问题仍是票价上涨和列车延误，WiFi改善并非首要诉求。
Oil was little changed in early Asian trading, as the US touted progress toward a peace deal with Iran despite fresh hostilities and uncertainty over the vital Strait of Hormuz.
中文摘要 亚洲早盘油价基本持平。美国方面表示在与伊朗达成和平协议上取得进展，但波斯湾地区紧张局势未消，霍尔木兹海峡通行安全的不确定性仍对油价构成影响。
Hands-on approach was viewed as aggressive by several colleagues at UK oil major
中文摘要 英国石油巨头BP解除董事长阿尔伯特·曼尼福德（Albert Manifold）的职务，起因是多名同事反映其管理风格强硬激进，并指控存在欺凌行为。曼尼福德的强势介入式管理风格引发内部不满，最终导致其离职。
Mubadala Investment Co., Abu Dhabi’s sovereign wealth fund, is seeking to raise $1.91 billion from an unregistered block sale of GlobalFoundries Inc. shares, according to people familiar with the matter.
中文摘要 阿布扎比主权财富基金穆巴达拉投资公司（Mubadala）正寻求通过非注册大宗交易出售GlobalFoundries股份，拟募资19.1亿美元。消息人士透露，此次为大宗协议转让，不经公开市场操作。
Kardigan Inc. filed for a US initial public offering that will help fund three late-stage drugs that target the root cause of cardiovascular diseases.
中文摘要 心血管疾病治疗公司Kardigan Inc.向美国提交IPO申请，募集资金将用于支持三款针对心血管疾病根本病因的后期临床阶段药物的研发。该公司目前有多个候选药物处于关键临床阶段。
Also in today’s newsletter: BP removes chair Albert Manifold, and Iran accuses US of ‘flagrant’ ceasefire violations
中文摘要 字节跳动向旗下人工智能团队提供特别股权激励，以应对竞争对手的人才挖角压力。此外，BP解除董事长曼尼福德职务，伊朗指责美国"公然"违反停火协议等多则要闻亦受关注。
Sam McKinney, KeyBanc Capital Markets vice president and equity research analyst, said that while US steel mills have mostly hedged rising electricity costs with long term contracts, they are impacted by increasing diesel prices to ship steel by truck and rail route. On the other side of the coin, s
中文摘要 KeyBanc资本市场副总裁、股票研究分析师萨姆·麦金尼表示，美国钢铁厂通过长期合同基本对冲了电力成本上涨风险，但柴油价格上涨推高了公路及铁路运输钢材的成本，战争与贸易协议双重因素正影响美国钢铁行业整体成本结构。
Enhanced Group Inc. shares fell to an all-time low after its inaugural sports-entertainment event was marred by technical problems and questions over its only world-record performance, disappointing investors weeks after its public listing.
中文摘要 Enhanced Group Inc.股价跌至历史低点，原因是其首场体育娱乐活动受技术故障困扰，且唯一一项世界纪录表现遭受质疑，令投资者大失所望。该公司在上市数周后即遭遇首秀危机。
The incredible shrinking stockmarket
中文摘要 《经济学人》分析指出，超大规模IPO（Giga-IPO）的涌现折射出公开股票市场面临的深层问题——上市公司数量持续萎缩，资本市场对企业的吸引力下降，巨型IPO现象在某种程度上是公开市场结构性困境的症状。
4 回复 · Apple 节点
42 回复 · Apple 节点
12 回复 · Apple 节点
6 回复 · Apple 节点
27 回复 · Apple 节点
4 回复 · Apple 节点
6 回复 · Apple 节点
4 回复 · Linux 节点
7 回复 · Python 节点
3 回复 · Linux 节点
之前一直在调整自己的Coding工作流，现在基本稳定成了一套组合： Cursor(Cursor++) + ACE + smart-search-cli + grill-me + Trellis + 其余MCP 之前使用的路径是： 孙佬的多模型协作 Workflow ↓ 风佬的 ccg ↓ Vibe-Skills ↓ 现在这套组合 1. 为什么我还在用Cursor 原因很简单：Cursor 的日常交互成本低 一边看代码； 一边看 diff； 一边让 Agent 查文件、改文件； 一边跑命令、看诊断； 一边整理文档或知识库。 Agent Window里具体编辑了什么也可以很清楚地看到。 2. Cu
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 以下为项目介绍正文内容，AI生成、润色内容已使用截图方式发出 一周开源总结 从上周开始开源无限画布，到现在大概过去了一周多的时间，GitHub收获了大概500多star。目前规划的是这个项目可以去接任意的xxx2api，比如chatgpt2api、grok2api、flow2ap
坐标山东三线小城 大学生朋友，被骗去“代销果酒“，然后被骗子演双簧直接骗走 300多 具体来说就是一个扮演上头经销，一个扮演顾客， 虽然不是很多，但是这个大学生本来就 只剩300多的生活费，这样一来更是雪上加霜 （我真无语了。。。咋就不能好好的搞点副业嘛现在有ai了大学生又闲赚点生活费还不容易了，偏偏想象有暴利的事情。其实现在暴利活有倒是有（ai不就是），可是做之前也不想想这一整条商业链路上创造价值的点在哪里，这钱凭啥让你挣了 ） 他还在那美滋滋的等那边给他回款呢，还好是没忍住和我说了说，我一提一看聊天记录才发现不对劲，这手法套路太典型了吧 于是我就提醒了他是被骗了 于是朋友打出了一个惊为天人
话不多说 先看总览 网络进你家，经过光猫，再到路由器，然后分给有线设备和无线设备。 这篇是硬件层次的，告诉你网络设备到底各自负责什么。 [网络知识扫盲]网络结构与 VPN / 代理服务区别 光猫 / Modem / ONT 先说光猫，也就是运营商送的那台。 传统意义上的 Modem 叫调制解调器，负责在模拟信号和数字信号之间转换。但现在很多地区已经换成了速度更快的光纤入户，所有我们一般将其叫做光猫。 名字什么的不太重要，这个就是用于连接到运营商网络的硬件入口。 不过随着时代的发展，现在运营商给出的光猫是一体机：它聚合了多个硬件的功能——光猫，拨号，做路由器，还提供 WIFI 网络。 这也是有人
就这个对话流 操作信息展示详细程度 能给你三档设置这一点细节就薄纱各路编辑器了 关于 2.1.20 上线的 collapsed read/search groups 你可以说 Agent Window 嗯 “抄” 的像 Codex Desktop 你信OpenAI对 Electron 的掌控实力还不如 信 Cursor 从 VSCode 那天然获得的性能优化 我只要想到他那坨架构 天天ab测试的开这个不开那个gate 我就三个头都大 某种程度上还不如 Claude Desktop 简单套个壳子来的清爽 更别说 Cursor checkpoint 和 多路模型可切换 上下文接续了 不再推进 Co
@sebastianjuly 这位老哥上手就来了一句让人摸不着头脑的发言，接着就跟自己互博，种种行为非常令人费解。 去查了一圈，很可能是AI养号，申请自述和加入缘由也是AI生成，直接送走。 这是第一例，绝不会是最后一例。 1 个帖子 - 1 位参与者 阅读完整话题
老项目的屎山代码很多，功能模块庞杂，让 codex 改代码偶尔会出现漏改，或动了本不该动的代码，所以整个项目需要先炼化后再 vibe coding. 使用 /goal 命令就很合适，把整个项目炼成 skills AGENT.md 再搭配手中的工作流 spec/superpower等等，codex再写起代码来能更精准，思考会更全面。 注意：需要开启 subagent 和 goal 功能,搭配 gpt-5.5 xhigh 效果更好，我试过用 5.4 也可以 但是炼化出的 skill 没 5.5 数量多和细节更完整。 /goal 你的任务是基于当前项目代码库，使用多路 subagent 和 $ski
大家再使用公益站链接codex 的时候应该都会跟我一样遇到一些小问题 无法使用codex中的插件功能，可以用codex++去解决：GitHub - b-nnett/codex-plusplus: Codex++ tweak system for the Codex desktop app · GitHub 我使用了但是发现mac的软件兼容性不太好，不太稳定 cc switch 切换key之后的历史记录丢失，可以用： GitHub - Dailin521/codex-provider-sync: Synchronize Codex session provider metadata across
{“key”:“sk-sGF2DjIfCTmteWvHcI8q6LmaCkJwtMlCjQvzBIRSBS7QmM5V”,“url”:“https://vip.lyclaude.site”} 感觉很快会挂，上成公益吧 110个plus，应该可以用一会，不封的话。 能看见key就是可以用，挂了我会删 限制了codex客户端可用，如果打测试可能不会同，直接接入cx用即可 网站已关注册和支付 服务器炸了，预计晚上九点恢复 来晚了，刚刚修好，已经恢复了 12 个帖子 - 9 位参与者 阅读完整话题
最近测试注册机搞出来的hotmail，大部分为长效，已经活了几天了 用cdk方式领 cdk.linux.do LINUX DO CDK Linux Do 社区 CDK 快速分享平台 - 让分享变得更简单 16 个帖子 - 14 位参与者 阅读完整话题