每日简报

2026-06-24

← 历史归档

calesthio/OpenMontage

Python · ★ 15,575 · 🍴 1,850 · 📈 3,592 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 开源 AI Agent 视频制作系统,集成 12 条流水线、52 种工具和 500+ 智能体技能,可将 AI 编程助手转变为完整的视频生产工作室,适合内容创作者和 AI 开发者快速生成视频内容。

ZhuLinsen/daily_stock_analysis

Python · ★ 47,005 · 🍴 42,367 · 📈 1,119 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,聚合多源行情与实时新闻,提供决策看板和自动推送,支持零成本定时运行,适合个人投资者和量化交易者进行日常分析。

mukul975/Anthropic-Cybersecurity-Skills

Python · ★ 19,660 · 🍴 2,288 · 📈 1,041 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 Agent 准备的 817 项结构化网络安全技能,映射到 MITRE ATT&CK、NIST CSF 2.0 等六大框架,遵循 agentskills.io 标准,适合安全团队训练智能体进行威胁检测与响应。

garrytan/gstack

TypeScript · ★ 114,040 · 🍴 16,883 · 📈 1,011 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 专属配置,包含 23 个自定工具,模拟 CEO、设计师、工程经理等角色,帮助开发者直接用 Claude Code 完成产品开发全流程。

bytedance/deer-flow

Python · ★ 73,883 · 🍴 9,967 · 📈 739 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 字节跳动开源的长时间跨度 SuperAgent 框架,支持研究、编码和创作,结合沙箱、记忆、工具、技能、子代理和消息网关,适用于构建多阶段复杂任务的 AI 智能体。

koala73/worldmonitor

TypeScript · ★ 59,070 · 🍴 9,283 · 📈 294 stars today

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

中文介绍 实时全球情报仪表盘,利用 AI 聚合新闻、监控地缘政治和基础设施动态,提供统一态势感知界面,适合分析师、研究员和决策者跟踪全球事件。

palmier-io/palmier-pro

Swift · ★ 8,400 · 🍴 551 · 📈 1,630 stars today

macOS video editor built for AI

中文介绍 专为 AI 设计的 macOS 视频编辑器,集成 AI 辅助功能,让用户在 Mac 上高效完成视频剪辑、特效添加等创作任务,适合视频博主和创意工作者。

anthropics/claude-plugins-official

Python · ★ 30,828 · 🍴 3,358 · 📈 77 stars today

Official, Anthropic-managed directory of high quality Claude Code Plugins.

中文介绍 Anthropic 官方维护的 Claude Code 插件目录,收录经过审核的高质量插件,便于开发者扩展 Claude 编码助手的能力,提升开发效率。

shanraisshan/claude-code-best-practice

HTML · ★ 59,418 · 🍴 5,981 · 📈 344 stars today

from vibe coding to agentic engineering - practice makes claude perfect

中文介绍 从「氛围编码」到「智能体工程」的实践指南,分享使用 Claude Code 的最佳方法,帮助开发者更专业地利用 AI 编码助手完成复杂项目。

revfactory/harness

HTML · ★ 7,434 · 🍴 1,045 · 📈 128 stars today

A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use.

中文介绍 一种元技能系统,能够设计领域特定的智能体团队、定义专业 Agent 并自动生成所需技能,适合需要搭建多 Agent 协作架构的开发者。

jamiepine/voicebox

TypeScript · ★ 33,114 · 🍴 3,996 · 📈 1,045 stars today

The open-source AI voice studio. Clone, dictate, create.

中文介绍 开源 AI 语音工作室,支持声音克隆、听写和创作,提供完整的语音合成与编辑功能,适用于配音、有声书制作和语音交互开发。

JCodesMore/ai-website-cloner-template

TypeScript · ★ 18,528 · 🍴 2,802 · 📈 826 stars today

Clone any website with one command using AI coding agents

中文介绍 利用 AI 编码 Agent 一键克隆任意网站,只需一条命令即可复制网站结构,适合快速搭建原型或复制现有站点的开发者。

byoungd/English-level-up-tips

★ 54,486 · 🍴 5,584 · 📈 125 stars today

An advanced guide to learn English which might benefit you a lot 🎉 . 人生进阶指南 离谱的人生 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 一份进阶英语学习指南,涵盖方法、资源和技巧,帮助学习者系统提升英语水平,适用于自学者和英语进阶用户。

DeusData/codebase-memory-mcp

C · ★ 12,865 · 🍴 933 · 📈 1,300 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知识图谱,支持 158 种语言,毫秒级查询,令牌消耗减少 99%,适合需要快速检索代码结构的开发者。

NousResearch/hermes-agent

Python · ★ 200,911 · 🍴 35,835 · 📈 936 stars today

The agent that grows with you

中文介绍 「随你成长」的 AI Agent,设计目标是与用户共同进化,逐步提升智能水平,但目前公开信息有限,可关注后续进展。

affaan-m/ECC

JavaScript · ★ 220,517 · 🍴 33,780 · 📈 593 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 Agent 框架性能优化系统,覆盖技能、直觉、记忆、安全等方面,采用研究优先的开发模式,兼容 Claude Code、Codex、Cursor 等主流编码 Agent,适合构建高性能智能体应用。

PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems

👍 78

LLM agents increasingly operate in large tool ecosystems, where real-world tasks require discovering relevant tools, inferring implicit sub-goals, and adapting to dynamic environments over long horizons. However, existing benchmarks rarely evaluate planning under retrieval-limited tool visibility. T

中文介绍 针对LLM工具智能体在大型工具生态中的长程规划评估不足,提出PlanBench-XL基准,重点测试在检索受限工具可视性下的规划能力。通过构建大规模工具集和长序列任务,量化智能体发现工具、推断子目标及适应动态环境的表现。

OpenRath: Session-Centered Runtime State for Agent Systems

👍 68

Modern agent systems often suffer from fragmented runtime state: transcripts, tool effects, memory events, workspace placement, branch provenance, and replay evidence are recorded separately and become difficult to inspect or reproduce. OpenRath addresses this issue with a PyTorch-like programming m

中文介绍 OpenRath采用PyTorch风格的编程模型,统一管理智能体系统的碎片化运行时状态(轨迹、工具效果、内存事件等),支持会话中心的断点续传与回放。解决了状态分割导致的难以复现和调试问题,提升了智能体系统的可观察性与可复现性。

DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams

👍 65

Massive unstructured multimodal streams suffer from high "data entropy," impeding both efficient human knowledge acquisition and high-quality AI post-training. Existing passive annotation paradigms, heavily reliant on heuristic rules or general VLMs, are costly, monotonous, and fail to unlock the de

中文介绍 针对海量无结构多模态流中数据熵高、被动标注成本高的问题,提出DataClaw0智能裁剪范式。通过智能体主动发现和定制化提取高价值多模态数据,降低标注成本并提升后训练数据质量。无需依赖启发式规则或通用VLM,实现自适应数据筛选。

EnterpriseClawBench: Benchmarking Agents from Real Workplace Sessions

👍 56

Enterprise agents increasingly operate inside workspaces: they read heterogeneous files, invoke tools, and deliver business artifacts. We introduce EnterpriseClawBench, an enterprise agent benchmark constructed from proprietary, real-world agent sessions. Starting from a large archive of workplace s

中文介绍 提出EnterpriseClawBench,基于真实企业工作日志构建的企业智能体基准。从大量工作场所会话中提取任务,包含异构文件读写、工具调用和业务制品生成。为评估企业智能体在真实动态环境中的表现提供标准化测试平台。

KaLM-Reranker-V1: Fast but Not Late Interaction for Compressed Document Reranking

👍 39

As retrieval systems scale, high-quality reranking becomes increasingly important. However, most existing rerankers, whether encoder-based or decoder-based, jointly encode the query and passage, tightly coupling their computation and limiting deployment efficiency as well as flexibility. We present

中文介绍 KaLM-Reranker-V1提出非交互式压缩文档重排序方法,将查询和文档分别编码后再计算相关性,摆脱了传统编码器或解码器联合编码的紧密耦合。在保持重排质量的同时大幅提升部署效率和灵活性,适用于大规模检索系统。

World Action Models: A Survey

👍 37

World Action Models (WAMs) are embodied predictive-action models that make a forecast of the future available to action. Recent WAMs repurpose large video generation models, and a parallel line relies on language or vision-language backbones without a video-generation core. This rapid expansion has

中文介绍 综述世界动作模型(WAM)——能够预测未来并指导行动的具身预测模型。梳理了基于大视频生成模型和基于语言/视觉语言主干的两条技术路线,总结其进展、挑战与应用场景,为后续研究提供系统性参考。

CLI-Universe: Towards Verifiable Task Synthesis Engine for Terminal Agents

👍 27

While recent LLM-based terminal agents have demonstrated promising capabilities, the scarcity of high-quality, executable training data remains a critical bottleneck. Existing synthesis pipelines typically scale by retrofitting surface-level artifacts into tasks, frequently yielding ambiguous instru

中文介绍 针对终端智能体训练数据稀缺、现有合成管道生成模糊指令的问题,提出CLI-Universe可验证任务合成引擎。通过结构化模板和可执行约束生成高质量、可验证的终端操作任务,解决了指令歧义和可执行性不足的瓶颈。

BioMatrix: Towards a Comprehensive Biological Foundation Model Spanning the Modality Matrix of Sequences, Structures, and Language

👍 19

We present BioMatrix, the first multimodal foundation model that natively integrates sequences, structures, and natural language for both molecules and proteins within a single decoder-only architecture. Existing biological foundation models pursue native multimodality and broad entity coverage sepa

中文介绍 BioMatrix是首个原生整合序列、结构和自然语言的多模态基础模型,采用单一解码器架构同时处理分子和蛋白质。克服了现有模型在原生多模态和广泛实体覆盖上的分离问题,实现跨模态全面融合。

HydraHead: From Head-Level Functional Heterogeneity to Specialized Attention Hybridization

👍 17

The quadratic complexity of attention poses a critical bottleneck for long-context processing, spurring interest in hybrid attention designs. Most open-source hybrid models adopt a layer-wise strategy. Yet, prior work has noted the inherent difficulty of integrating Linear Attention (LA) with Full A

中文介绍 针对长上下文处理中注意力二次复杂度瓶颈,提出HydraHead头级功能异构混合注意力。不同于分层策略,在头级别混合线性注意力与全注意力,克服两者集成困难,在保持效率的同时提升长序列建模能力。

SkillHarness: Harnessing Safe Skills for Computer-Use Agents

👍 16

Computer-Use Agents (CUAs) are increasingly deployed in dynamic interactive environments, creating a growing need for continual skill learning during interaction. Recent approaches address this challenge by learning reusable skills from successful trajectories. However, these skill learning methods

中文介绍 SkillHarness为计算机使用智能体提供安全技能学习框架,从成功轨迹中提取可复用技能,并确保技能执行的安全性。在动态交互环境中实现持续学习,避免不安全行为,提升智能体在实际部署中的可靠性。

Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding

👍 15

Autoregressive generation in large language models (LLMs) conventionally decodes from the final layer, assuming that deeper representations yield more reliable next-token predictions. We revisit this assumption by revealing a recurring Guess-Refine-Perturb dynamic: early layers form coarse guesses,

中文介绍 发现LLM自回归生成中「猜测-修正-扰动」的动态:深层并非总是最优。提出Confident Layer Decoding,从浅层或中层选取置信度高的解码层,而非固定使用最后一层。有效缓解对齐税,在多个任务上提升生成质量。

Learning from Your Own Mistakes: Constructing Learnable Micro-Reflective Trajectories for Self-Distillation

👍 15

Self-distillation improves reasoning in large language models by using the model's own rollouts as training signal, typically through implicit logit-level alignment that minimizes KL divergence toward a privileged target distribution. However, because this supervision is generated via uncontrolled s

中文介绍 提出可学习微反思轨迹自蒸馏方法,替代传统隐层对齐。让模型从自身错误输出构建反思轨迹,通过有监督学习纠正错误,避免不可控采样导致的噪声。在推理任务上提升自蒸馏效果,无需外部教师。

Unlimited OCR Works

👍 13

Recently, end-to-end OCR models, exemplified by DeepSeek OCR, have once again thrust OCR into the spotlight. A widely held view is that employing a large language model (LLM) as the decoder allows the model to leverage the prior distribution of language, leading to improved OCR performance. However,

中文介绍 挑战「LLM解码器利用语言先验提升OCR」的普遍观点,证明端到端OCR模型性能主要得益于更好的视觉编码而非语言模型。通过对比实验揭示先验分布的潜在弊端,为OCR模型设计提供新视角。

Foresight: Failure Detection for Long-Horizon Robotic Manipulation with Action-Conditioned World Model Latents

👍 11

Long-horizon tasks are common in real-world robotic deployments, yet failure detection for such tasks remains underexplored. Detecting failures in long-horizon robotic tasks is particularly challenging because failure onset is often ambiguous and dense temporal annotations are typically unavailable.

中文介绍 Foresight利用动作条件世界模型潜在变量,在长时程机器人操作任务中检测早期故障。无需密集时间标注,通过预测未来状态与预期差异识别异常,在真实机器人实验中有效提前发现失败趋势。

Self-Compacting Language Model Agents

👍 11

Long agent traces composed of chains of thought and tool calls accumulate stale content that anchor subsequent generations, and eventually outgrow the context window. Existing scaffolds mitigate it with fixed-interval compaction triggered at a token threshold. Such triggers pay no heed to trajectory

中文介绍 针对智能体长轨迹中陈旧内容累积导致上下文窗口溢出问题,提出自压缩机制。智能体根据轨迹语义动态决定何时压缩,而非固定token阈值。在保持关键信息的同时有效压缩上下文,提升长任务执行效率。

Notes2Skills: From Lab Notebooks to Certainty-Aware Scientific Agent Skills

👍 9

Scientific discovery workflows usually contain and rely heavily on lab notes, where researchers record observations, interpret uncertain results, and plan follow-up experiments. Such informative lab notes preserve evolving scientific reasoning and author uncertainty, rather than polished final resul

中文介绍 利用实验记录本中富含的不确定推理,提出Notes2Skills将非结构化笔记转化为带有置信度的科学智能体技能。保留原始不确定性信息,使智能体在规划后续实验时更可靠,适用于真实科学发现流程。

Training Open Models for Agentic Phone Use

👍 9

Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while s

中文介绍 针对手机智能体训练难题(真实设备慢、状态复杂、难以重置),提出使用模拟器和真实设备混合训练策略。训练开放模型在真实手机上执行拨号、发消息等代理任务,在多个基准上取得与闭源模型相当的性能。

Tmax: A simple recipe for terminal agents

👍 8

Terminal-using agents have quickly become the most popular downstream application of language models (LMs). Despite their prevalence, relatively little academic work has examined RL-based training of these models, likely due to difficult benchmarks, a lack of data, and a lack of simple baseline reci

中文介绍 Tmax提出基于强化学习的终端智能体训练简易配方,解决因基准复杂、数据匮乏导致的训练困难。通过精心设计的奖励函数和模拟环境,在终端任务上取得显著提升,为后续研究提供可复现的基线。

Toward Open Weight Models Without Risks: Separating Public and Private Capabilities in LLMs

👍 7

Open-weight Large Language Models (LLMs) enable scientific progress and broad deployment. However, they make it difficult to control access to sensitive capabilities. Current practice either suppresses dangerous capabilities before release or mediates access through closed services that use speciali

中文介绍 针对开源大模型难以控制敏感能力的问题,提出参数分离方案:将公共能力和敏感能力拆解到不同模块,仅公开发布公共部分。通过激活特定参数子集实现能力隔离,既保证开源透明性,又防止危险能力滥用。

PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning

👍 7

Latent action pretraining learns representations of visual change from pairs of observations, but existing methods typically encode each transition as a single unstructured representation that entangles transition extent and transition mode. We introduce Polar Latent Actions with Radial structure (P

中文介绍 PoLAR提出径向结构的极化潜在动作,将机器人策略学习中的过渡幅度和过渡模式解耦为独立因子。相比传统单一潜在表征,更精细地建模视觉变化,提升策略学习效率和在多样化任务中的泛化能力。

40 Claude Opus 4.8 Workflows That Make Money While You Sleep

@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,重点不在于回答问题,而是让模型自动运行商业任务赚取收入。少数人已实现「睡觉时赚钱」,关键在于工作流设计而非模型本身。

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 Engram发布,目标是构建能学习用户并深度理解其工作的AI模型。强调在上下文层面扩展计算能力,让模型真正理解用户的领域知识,而非仅依赖训练数据。

ORACLE: Official AI Agents Trade on Polymarket

@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推出官方AI代理,可在Polymarket预测市场进行交易。自主AI代理策略已占Polymarket 30%以上活动,成为2026年预测市场最有效方式之一。

ORACLE: Official AI Agents Trade on Polymarket

@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推出官方AI代理,用于Polymarket预测市场交易。自主AI代理策略占据市场30%以上活动,成为2026年预测市场主流方式。

Interactions API: Our primary interface for Gemini models and agents

@GoogleAIStudio · 179.0K 粉丝 · 138.2K 阅 · 504 赞 · 42 转

Today we're announcing that the Interactions API has reached general availability and is now our primary API for interacting with Gemini models and agents. We launched its public beta in December

中文介绍 Google AI Studio宣布Interactions API正式可用,成为与Gemini模型和代理交互的主要接口。该API自12月公测后正式发布,统一了对话与应用开发流程。

ORACLE: Official AI Agents Trade on Polymarket

@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推出官方AI代理,用于Polymarket预测市场交易。自主AI代理策略占据市场30%以上活动,成为2026年预测市场最有效方式之一。

How To Use Loop Engineering To Build A Self-Improving Quant Trading System

@RohOnChain · 51.4K 粉丝 · 108.3K 阅 · 501 赞 · 65 转

I will break down exactly how to build the loops that run an entire quant trading system on their own. Let's get straight to it. Bookmark This - I'm Roan, a backend developer working on system

中文介绍 详细讲解如何使用循环工程构建自我改进的量化交易系统。作者RohOnChain分解了让系统自动运行循环的关键步骤,强调通过循环设计而非静态提示实现自适应交易。

How we built the world’s fastest API for GLM-5.2

@philipkiely · 8.6K 粉丝 · 73.1K 阅 · 544 赞 · 36 转

GLM-5.2 is the biggest news in open models since DeepSeek-R1. It’s easy to see why. GLM-5.2 delivers comparable performance to GPT 5.5 and Opus 4.8 at a fraction of the cost, generally 70-80% less

中文介绍 介绍GLM-5.2开源模型,性能对标GPT 5.5和Opus 4.8,成本降低70-80%。作者团队为其构建了全球最快API,分享技术实现细节。

The Next 5 Years: A Supersonic Tsunami

@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 Peter Diamandis引用Elon Musk将未来五年形容为「超音速海啸」——变化速度极快,让人来不及反应。探讨技术加速对商业和社会的冲击。

Loop Engineering Clearly Explained

@akshay_pachaar · 278.2K 粉丝 · 44.0K 阅 · 504 赞 · 72 转

Half your feed is suddenly saying the same thing. Stop prompting your agents, start engineering loops. Boris Cherny, the person who built Claude Code, said it plainly: "I don't prompt Claude anymore.

中文介绍 清晰解释循环工程概念,引用Claude Code创造者Boris Cherny的观点:不再提示agent,而是设计循环。说明循环工程如何替代传统prompt方式,提升agent自主性。

How GPT-5 helped immunologist Derya Unutmaz solve a 3-year-old mystery

GPT-5 Pro helped solve a 3-year-old immunology mystery, offering insights into T cell behavior. The breakthrough could support cancer and autoimmune research.

中文介绍 GPT-5 Pro帮助免疫学家Derya Unutmaz解决了一个持续三年的免疫学谜题,揭示了T细胞行为的新见解。这一突破可能支持癌症和自身免疫疾病研究。

Helping build shared standards for advanced AI

OpenAI helps build shared standards for advanced AI, supporting evaluation frameworks, safety practices, and global cooperation through the Appia Foundation.

中文介绍 OpenAI通过Appia基金会参与构建高级AI共享标准,支持评估框架、安全实践和全球合作。

The $400 million machine powering the future of chipmaking

Jos Benschop is climbing a ladder to get to the top of his newest machine. It’s a bit of a schlep. The contraption is the size of a double-decker bus—more than 150 tons of gleaming precision-milled aluminum covered in thousands of snaking tubes, colored cables, and pressurized tanks. From the ground

中文介绍 ASML价值4亿美元的芯片制造机器,大小如双层巴士,重150多吨,由精密铣削铝材制成,内部布满管线,被认为是芯片制造未来的关键设备。

[AINews] SpaceX is already a $28B/yr Neocloud

a quiet day lets us reflect on some numbers from Jamin Ball.

中文介绍 据Jamin Ball数据,SpaceX已发展成年收入280亿美元的Neocloud服务商。

How Omio is building the future of conversational travel

Discover how Omio uses OpenAI to power conversational travel experiences, accelerate product development, and transform into an AI-native company.

中文介绍 Omio利用OpenAI技术打造对话式旅行体验,加速产品开发,转型为AI原生公司。

Red-Teaming after Mythos — Zico Kolter & Matt Fredrikson, Gray Swan

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讨论AI安全,指出AI安全不仅仅是“AI网络安全”。

Three things to watch amid Anthropic’s latest feud with the government

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与美国政府最新冲突中,有三个关键点值得关注。

Patch the Planet: a Daybreak initiative to support open source maintainers

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推出Patch the Planet,属于Daybreak计划,帮助开源维护者利用AI和专家评审发现、验证和修复漏洞。

Daybreak: Tools for securing every organization in the world

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工具集,包括Codex Security和GPT-5.5-Cyber,帮助组织大规模发现、验证和修补漏洞。

not much happened today

**OpenAI** expanded its **Daybreak** program with the **GPT-5.5-Cyber** model, focusing on closed-loop patch generation for cybersecurity, scanning over 30 million commits and covering major projects like cURL and Python. The release sparked debate on policy and export controls, contrasting with **A

中文介绍 OpenAI扩展Daybreak计划,发布GPT-5.5-Cyber模型,专注于网络安全闭环补丁生成,已扫描超3000万次提交,覆盖cURL和Python等主要项目,引发政策与出口管制讨论。

市场总览

今日市场整体承压,加密恐慌贪婪指数仅为17(极度恐慌),加密总市值2.23T美元(24h -2.08%),BTC主导率56.2%,ETH占9%。美股方面,主要指数和科技股普遍下跌:QQQ跌3.29%,MSFT、NVDA、TSLA跌幅较大,多数个股技术面呈空头排列或接近超卖。中概股全面弱势,BABA、PDD等RSI均进入超卖区域,价格接近52周低点。商品期货方面,WTI原油跌近10%至72.7,RSI 27.9超卖;黄金亦下跌至4115.6,MACD死叉。外汇方面,美元指数DXY走强至101.39,RSI 75.3超买且多头排列,美元/人民币则平稳。VIX恐慌指数上升12.79%至19.49,MACD金叉但整体趋势中性。整体看,风险资产普遍偏弱,避险情绪有所升温。

今日关注

DX-Y.NYB 美元指数 DXY
偏上行

当前价格101.39,高于所有均线(SMA20 99.79, SMA50 99.02, SMA200 98.71),形成多头排列;RSI14 75.3 进入超买区,MACD 0.4899 大于信号线0.3447 维持金叉,显示短期动能强劲但超买需警惕。

BABA 阿里巴巴 (BABA)
偏下行

当前价格102.6,远低于各均线(SMA20 118.22, SMA50 128.35, SMA200 148.89),空头排列明显;RSI14 21.2 处于超卖区域,MACD -7.0192 在信号线 -5.5618 下方呈死叉,近5日跌8.84%,距52周低仅0.75%,技术面极度弱势。

^VIX VIX 恐慌指数
中性

当前19.49,高于SMA20 17.58和SMA50 17.76,但低于SMA200 18.59,趋势中性;RSI14 54.4 正常,MACD 0.025 上穿信号线-0.05 形成金叉,近5日上涨20.31%,反映市场恐慌情绪上升但尚未极端。

全部资产

^VIX

VIX 恐慌指数

$19.49 +12.79%
5 日
+20.31%
距 52w 高
-44.8%
RSI(14)
54.4
趋势
中性
SMA 20 / 50 / 200
17.58 / 17.76 / 18.59
MACD / 信号
0.025 / -0.050
死叉(SMA50↓SMA200) (8 天前)MACD 金叉 (今天)

^TNX

10Y 美债收益率 (%)

$4.49 +0.94%
5 日
+0.13%
距 52w 高
-10.1%
RSI(14)
52.2
趋势
多头
SMA 20 / 50 / 200
4.49 / 4.43 / 4.22
MACD / 信号
0.005 / 0.014
多头排列

DX-Y.NYB

美元指数 DXY

$101.39 +0.37%
5 日
+1.77%
距 52w 高
-0.0%
RSI(14)
75.3
趋势
多头
SMA 20 / 50 / 200
99.79 / 99.02 / 98.71
MACD / 信号
0.490 / 0.345
MACD 金叉 (3 天前)RSI 超买接近 52 周高多头排列

SPY

S&P 500 ETF

$733.58 -1.45%
5 日
-2.82%
距 52w 高
-3.5%
RSI(14)
45.9
趋势
多头
SMA 20 / 50 / 200
746.56 / 732.03 / 689.29
MACD / 信号
2.466 / 4.607
多头排列

QQQ

Nasdaq 100 ETF

$713.65 -3.29%
5 日
-4.08%
距 52w 高
-4.7%
RSI(14)
48.8
趋势
多头
SMA 20 / 50 / 200
727.85 / 697.79 / 630.16
MACD / 信号
7.906 / 10.255
多头排列

AAPL

Apple

$294.30 -0.91%
5 日
-0.72%
距 52w 高
-7.3%
RSI(14)
47.1
趋势
多头
SMA 20 / 50 / 200
302.27 / 290.15 / 268.76
MACD / 信号
0.578 / 2.329
多头排列

MSFT

Microsoft

$373.94 +1.80%
5 日
-6.46%
距 52w 高
-32.7%
RSI(14)
35.3
趋势
空头
SMA 20 / 50 / 200
408.33 / 412.92 / 449.99
MACD / 信号
-11.251 / -6.188
空头排列

NVDA

Nvidia

$200.04 -4.13%
5 日
-5.84%
距 52w 高
-15.4%
RSI(14)
42.5
趋势
中性
SMA 20 / 50 / 200
210.49 / 210.03 / 190.23
MACD / 信号
-1.684 / -0.575

GOOGL

Alphabet

$346.13 -1.02%
5 日
-6.29%
距 52w 高
-15.3%
RSI(14)
37.4
趋势
中性
SMA 20 / 50 / 200
367.89 / 368.58 / 312.26
MACD / 信号
-4.400 / -1.852

TSLA

Tesla

$381.61 -5.79%
5 日
-7.18%
距 52w 高
-23.5%
RSI(14)
40.9
趋势
空头
SMA 20 / 50 / 200
410.84 / 404.33 / 417.53
MACD / 信号
-4.306 / -1.606
空头排列

META

Meta

$562.20 -0.29%
5 日
-5.27%
距 52w 高
-29.4%
RSI(14)
39.0
趋势
空头
SMA 20 / 50 / 200
594.90 / 619.25 / 653.12
MACD / 信号
-13.241 / -10.953
空头排列
加密恐慌贪婪
17
极度恐慌
加密总市值
$2.23 T
-2.08% / 24h
BTC 主导率
56.2%
ETH 9.0%
24h 成交量
$73.6 B
活跃币 17,408

BTC-USD

Bitcoin

$62,768.08 -1.85%
5 日
-0.20%
距 52w 高
-50.3%
RSI(14)
37.5
趋势
空头
SMA 20 / 50 / 200
63,459.71 / 71,465.72 / 76,449.67
MACD / 信号
-1,968.871 / -2,364.355
空头排列

ETH-USD

Ethereum

$1,669.90 -3.28%
5 日
-2.32%
距 52w 高
-66.3%
RSI(14)
37.4
趋势
空头
SMA 20 / 50 / 200
1,694.49 / 1,965.50 / 2,349.97
MACD / 信号
-65.168 / -81.421
空头排列

SOL-USD

Solana

$69.79 -2.95%
5 日
+0.23%
距 52w 高
-72.4%
RSI(14)
43.3
趋势
空头
SMA 20 / 50 / 200
68.76 / 79.22 / 96.75
MACD / 信号
-1.715 / -2.577
空头排列

BABA

阿里巴巴 (BABA)

$102.60 -2.26%
5 日
-8.84%
距 52w 高
-46.7%
RSI(14)
21.2
趋势
空头
SMA 20 / 50 / 200
118.22 / 128.35 / 148.89
MACD / 信号
-7.019 / -5.562
RSI 超卖接近 52 周低空头排列

PDD

拼多多 (PDD)

$76.56 -1.98%
5 日
-8.37%
距 52w 高
-45.1%
RSI(14)
27.9
趋势
空头
SMA 20 / 50 / 200
83.55 / 92.84 / 110.12
MACD / 信号
-4.266 / -4.031
RSI 超卖接近 52 周低空头排列

JD

京东 (JD)

$26.12 -3.33%
5 日
-8.93%
距 52w 高
-29.1%
RSI(14)
28.1
趋势
空头
SMA 20 / 50 / 200
28.63 / 29.98 / 30.20
MACD / 信号
-0.847 / -0.612
RSI 超卖空头排列

0700.HK

腾讯控股 (0700.HK)

HK$414.80 -4.20%
5 日
-9.75%
距 52w 高
-39.3%
RSI(14)
34.0
趋势
空头
SMA 20 / 50 / 200
447.41 / 465.97 / 564.14
MACD / 信号
-8.007 / -5.743
MACD 死叉 (1 天前)接近 52 周低空头排列

GC=F

黄金期货

$4,115.60 -1.59%
5 日
-4.91%
距 52w 高
-26.3%
RSI(14)
34.0
趋势
中性
SMA 20 / 50 / 200
4,338.52 / 4,533.17 / 4,442.11
MACD / 信号
-101.229 / -94.325
MACD 死叉 (1 天前)

CL=F

WTI 原油期货

$72.70 -2.83%
5 日
-9.97%
距 52w 高
-39.2%
RSI(14)
27.9
趋势
中性
SMA 20 / 50 / 200
86.21 / 93.38 / 73.77
MACD / 信号
-5.710 / -4.126
RSI 超卖

USDCNY=X

美元 / 人民币

¥6.79 +0.22%
5 日
+0.48%
距 52w 高
-5.9%
RSI(14)
56.5
趋势
空头
SMA 20 / 50 / 200
6.77 / 6.80 / 6.95
MACD / 信号
-0.007 / -0.011
接近 52 周低空头排列
风险提示

过去走势不代表未来表现,本报告仅基于当前公开行情数据计算的技术指标进行客观状态描述,不构成任何投资建议。仅供技术指标解读参考。

U.S. and Iran Offer Conflicting Accounts of Nuclear Discussions

President Trump said Iran had agreed to the “highest level” inspections, hours after an Iranian official said there were “no detailed discussions on the nuclear issue,” as the two sides continued to present different narratives of their latest talks.

中文摘要 美伊就核谈判各执一词。特朗普称伊朗同意接受最高级别核查,伊朗官员则否认有详细核讨论,双方对最新会谈说法矛盾。

Australia politics live: Energy minister asks regulator to investigate big jump in power companies’ supply fees

Social services minister says Pauline Hanson ‘completely unsympathetic’ to financial stress of having a baby. Follow today’s news live Get our breaking news email, free app or daily news podcast Ruston pushes for overhaul to controversial aged care tool Over to her portfolio, the shadow health and a

中文摘要 澳大利亚能源部长要求监管机构调查电力公司供应费大幅上涨。社会服务部长批评参议员Pauline Hanson对生育财务压力缺乏同情心。

Protesters sentenced to decades in US prison over alleged antifa ties

Civil liberties advocates have expressed alarm over the case involving a protest outside Prairieland Detention Center.

中文摘要 美国多名抗议者因涉嫌与反法西斯组织有联系被判处数十年监禁。民权倡导者对Prairieland拘留中心外抗议案判决表示担忧。

Stanford was their golden ticket - could AI help or hinder that?

The BBC spoke with Stanford University graduates about what they really think about artificial intelligence.

中文摘要 BBC采访斯坦福大学毕业生,了解他们对人工智能的看法,探讨AI是帮助还是阻碍其发展。

Multiple Airbus A380 planes to be grounded for urgent inspections after cracks found in wings

Airbus to inspect 15 Emirates and one Qantas plane after cracks found in structural wing beam of some jets Follow our Australia news live blog for latest updates Get our breaking news email, free app or daily news podcast Airbus will inspect 16 A380 planes, five of them immediately, after cracks wer

中文摘要 因机翼结构梁发现裂纹,空中客车将对16架A380飞机进行紧急检查,其中5架立即停飞,涉及阿联酋航空15架和澳洲航空1架。

US Supreme Court says Rastafarian man shaved by prison guards can’t sue

The high court has upheld ruling that prisoners cannot sue prison staff for money damages under religious liberties law.

中文摘要 美国最高法院裁定,一名拉斯塔法里教徒囚犯被狱警剃须后不能起诉要求经济赔偿,理由是根据宗教自由法囚犯不能起诉。

Israel’s deliberate targeting of Palestinian children

New UN report documents the toll of Israel's war on Gaza and its occupation of the West Bank on Palestinian children.

中文摘要 联合国新报告记录以色列在加沙战争和西岸占领中针对巴勒斯坦儿童的行为,指出其故意造成伤害。

Scientists find ‘smoking gun’ evidence of world’s oldest meteorite strike in Western Australia

Curtin University researchers use innovative techniques to date three-billion-year-old impact crater in Pilbara region A meteorite that struck Earth three billion years ago left behind a “smoking gun” – evidence of the world’s oldest impact crater in a remote part of Australia. Ancient rocks in West

中文摘要 科廷大学研究人员在西澳大利亚皮尔巴拉地区发现30亿年前陨石撞击坑,确认为世界最古老陨石撞击证据。

Paraguay’s Almiron suspended for one game after red card for covering mouth

FIFA also stripped journalist Jorge Chipi Vera of World Cup credentials after he criticised Almiron's red card on air.

中文摘要 巴拉圭球员阿尔米隆因比赛中捂嘴动作被红牌罚下,停赛一场。国际足联还撤销了批评此判罚的记者Jorge Chipi Vera的世界杯采访证。

U.N. Report Says Israeli Killings of Gaza Children Post-Truce Amount to Genocide

Israel’s U.N. mission dismissed the report as a “libelous sham,” and called the investigating commission a “fundamentally flawed mechanism.”

中文摘要 联合国报告称以色列在停火后杀害加沙儿童的行为构成种族灭绝。以色列代表团驳斥为“诽谤性骗局”。

Gold Drops Below $4,100 as Tech-Led Selloff Spurs Liquidation

Gold extended a decline, as a tech-led selloff on Wall Street prompted investors to cut bullion holdings to cover losses elsewhere in their portfolios.

中文摘要 黄金价格跌破4100美元关口,华尔街科技股领跌促使投资者减持黄金以弥补其他投资组合的损失,金价延续跌势。

Google's YouTube settles social media addiction case with teen

The 15-year-old plaintiff's case against three other tech giants begins next month.

中文摘要 YouTube与一名15岁青少年原告就社交媒体成瘾案达成和解,该原告对其他三家科技巨头的诉讼将于下月开庭。

UK Issues Energy Supply Warning as Heat Wave Strains Power Grid

The UK’s grid operator issued a rare summer power supply warning for Wednesday evening as soaring temperatures put stress on the energy system.

中文摘要 英国电网运营商罕见发布夏季电力供应预警,高温热浪导致能源系统承压,周三晚间可能面临供电紧张。

Stanford was their golden ticket - could AI help or hinder that?

The BBC spoke with Stanford University graduates about what they really think about artificial intelligence.

中文摘要 BBC采访斯坦福大学毕业生,了解他们对人工智能的看法:AI究竟是助力还是阻碍了他们的“黄金门票”之路。

Ten years on, Brexit's economic impact is becoming clearer

A decade ago, many economists argued the UK would sustain longer-term economic damage by leaving the EU. So what did happen?

中文摘要 英国脱欧十周年,众多经济学家此前预言的长期经济损害逐步显现,其经济影响正变得更加清晰。

The economic challenges facing the next prime minister

Though the person in charge of the country will change, the fiscal issues remain the same.

中文摘要 下任英国首相将面临与现任相同的财政挑战,尽管领导人变更,但经济难题依旧未变。

Xi Pressures Takaichi by Throttling Key Mineral Exports to Japan

China is choking shipments of some critical minerals to Japan, a slowdown that’s hurting companies and prompting calls for Prime Minister Sanae Takaichi to find a diplomatic off-ramp with Beijing.

中文摘要 中国限制对日本出口关键矿物,对日本企业造成冲击,外界呼吁首相高市早苗寻求外交途径缓和与北京的关系。

UAE Oil Exports Surge to 85% of Pre-War Levels, IEA Says

Oil exports from the United Arab Emirates in early June recovered to nearly 85% of the country’s pre-Iran war levels — rebounding even before Washington and Tehran inked an interim peace deal as the Gulf nation drew on pipelines, storage and alternate shipping routes, according to a report from the

中文摘要 国际能源署称,阿联酋6月初石油出口已恢复至伊朗战争前水平的85%,通过管道、储油和替代航线实现反弹。

Asian Stocks Set to Extend AI Chip-Fueled Selloff: Markets Wrap

Stocks in Asia looked poised to extend Tuesday’s losses after a bruising Wall Street tech-led selloff deepened concerns that the artificial intelligence-driven equity rally may have run too far.

中文摘要 亚洲股市预计延续跌势,华尔街科技股遭抛售加剧了市场对人工智能驱动涨势可能过度的担忧。

Oil Holds Drop as More Tankers Cross Hormuz After Peace Talks

Oil held declines as more tankers openly cross the Strait of Hormuz, with the US and Iran making progress toward ending the war.

中文摘要 油价继续下跌,美伊和平谈判取得进展,更多油轮公开穿越霍尔木兹海峡,市场预期战争将结束。

CC Enable Computer Use

从求助官方claude code cli如何使用computer use继续讨论: 有人问了 花了点时间看了下 Patch 了开关 补了 两个操作库 成功启用 Computer Use @cometix/claude-code 2.1.187 开始补齐该能力 apply-claude-code-computer-use-fix.zip (6.6 KB) // 开关配置 : computerUseEnabled true 才会注入启用MCP "computerUseEnabled": true, "computerUseConfig": { "pixelValidation": false, "

微信小程序mmtls协议逆向分析小记

这次打算好好水一篇文章,我们就来慢慢的唠! 前言 什么是mmtls 我们来看看腾讯安全自己的介绍 简单点说,mmtls就是腾讯自己魔改的私有协议,最早就要追溯到17年了,微信当时的技术团队自研魔改的。 在手机微信上抓包的佬应该会经常见到。 小程序的mmtls有啥用 mmtls在小程序的使用途径就是云函数,最常用的,也是我们这次搞定的对象 wx.login() 小程序的官方文档 我们在打开小程序的时候,会有一个快捷登录,又或者没有登录,直接打开就发现小程序已经登录账号了 这都是wx.login进行返回的code,有了这个code,就可以登录小程序,跟小程序自己写的后端换token 又或者是霸王茶

2026.6.23两大ai巨头炸

oai和claude好嗑啊 openrouter紧随其后 41 个帖子 - 40 位参与者 阅读完整话题

舆论对立挑不起来,跟我玩栽赃是吧

热更新 107.173.42.94 分割线 从 cpa 被来自 racknerd 的ip 107.173.42.94 盗刷, 举报有效, 已经suspend了该server. - #15,来自 cheeryman 继续讨论, 我的基本上所有的服务器都在ovh,都是物理机,我不至于穷到用rn,别瞎说 有一点家宽和线路机 具体请看,君の守望 47 个帖子 - 42 位参与者 阅读完整话题

「裸感」衣服的原理?|一位百合女的旧话重提

[!quote]为什么会有nsfw呢 故事背景:吃饭前把这个话题要发的正文写好了,然后就去吃饭了,期间一直在想这个话题 直到现在还是不知道该从哪一件事说起 ) 其他内容可以参考历史文献: 在另一个话题中 你怎么知道我作为女同超喜欢舔自己皮肤的? ) 手心更是一般不考虑 ) 事实上,手虽然是日常生活中最无可置疑的主角之一,但手背也没接触过什么东西,按理说应该也挺干净的 洗完手之后香香的 所以君子动手不动嘴(除非有时候忍不住),我一般还是以到处摸来摸去(核心是滑滑梯版触感 ),或者是贴贴脸颊,贴贴额头之类的 贴鼻子这个在二次元里还是蛮浪漫的,但是现实中试的话只会让鼻子痛痛的() 所以终于引出了本次

各位佬,千万别做手术了。

原来还想着自己一个人做完手术(前交叉韧带撕裂),自己回去得了,不给任何人添麻烦(家里还有一妻一女),没想到医院的做手术必须要有人陪。医生告诉我,你麻醉完,做完手术就是一个废人,吃喝拉撒全部在床上。 70 个帖子 - 51 位参与者 阅读完整话题

明天四点,准备飞升三级

一直很佛系,不过最近token有点紧张,君佬的公益站又有等级限制,对升三级燃起了激情哈哈哈哈哈。因为自己还在学习,很多领域都一知半解所以不敢发言。Anyway,明早四点就要飞升三级,l站邮箱,君佬公益站,更多精彩内容还是很期待的。努力学习,持续学习。 68 个帖子 - 56 位参与者 阅读完整话题

盘点目前可用的GLM-5.2渠道

GLM-5.2 都发布 10 天了我还没抢到 Coding Plan,我已急哭……不过好消息是它是开源的,目前市面上已有多种第三方渠道接入。以下盘点截至目前 GLM-5.2 的可用渠道(包含我已实测的和道听途说的)。 省流:付费订阅 OpenCode Go 综合最佳;免费尝鲜可以选官方 ZCode 或 阿里云百炼。 注:如果怀疑渠道掺水,安全边界也不失为一个检查的好办法。但注意即便是没有内容审查(中途截断 / 无法发送)的渠道,GLM 自身也可能因为安全对齐而说抱歉(但此时输出是完整的)。 1. 官方渠道价格参考(作为基线) 1.1 官方 API(按量付费) 模型文档:GLM-5.2 - 智谱

你以为去了寺庙,就能脱离世俗?

久违了佬友们,算下时间已经有好几个月没发过帖子了。 时隔几个月重新发帖,主要是想跟佬友们一起分享我近期的感悟与心得。 前段时间身心俱疲,做什么都提不起劲,不仅耽误了工作,也影响了身边的人际关系。 于是就约几个朋友一起吃饭谈谈心,饭桌上有个朋友听了我的状态后,便推荐我去山里的寺庙住一段时间,说他之前状态跟我一样,后来去了山里跟师父们做做功课,念念经,喝喝茶,回来后整个人都充满了能量。 我听了他的描述后,脑海顿时有了画面,青灯古佛,晨钟暮鼓,在香火旁念着梵音的修行人,光想着就觉得轻松了些。 饭后我和朋友要了他去的寺庙地址,打算抽空去挂个单清净一下。 估计有佬友不知道什么是挂单,挂单指的是云游的僧人