2026-05-31

microsoft/markitdown

Python · ★ 132,570 · 🍴 9,072 · 📈 2,470 stars today

Python tool for converting files and office documents to Markdown.

中文介绍 微软开源的 Python 工具，可将各类文档和 Office 文件（如 Word、Excel、PDF）转换为 Markdown 格式，便于内容提取和后续处理。适用于文档迁移、知识库构建等场景，可集成到自动化工作流中。

harry0703/MoneyPrinterTurbo

Python · ★ 72,107 · 🍴 10,336 · 📈 2,768 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 大模型的一键生成高清短视频工具。用户只需输入主题或文案，即可自动合成带配音、字幕和背景音乐的短视频，适合社交媒体内容创作、营销推广等快速产出场景。

anthropics/claude-code

Python · ★ 128,429 · 🍴 20,954 · 📈 592 stars today

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

中文介绍 Claude 的智能编码代理，运行于终端中，能理解代码库并协助完成日常任务、解释复杂代码、管理 git 工作流。适用于开发者提高编码效率，尤其是在大型项目中进行快速迭代和调试。

cursor/plugins

TypeScript · ★ 1,473 · 🍴 118 · 📈 205 stars today

Cursor plugin specification and official plugins

中文介绍 Cursor 编辑器的插件规范及官方插件集合。通过该仓库，开发者可以了解如何扩展 Cursor 功能，并获取官方维护的插件，增强代码编辑体验。适用于需要定制化编辑器的用户。

revfactory/harness

HTML · ★ 4,283 · 🍴 627 · 📈 55 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，用于设计领域特定的智能体团队、定义专业化代理并生成它们使用的技能。适合开发多智能体协作的复杂任务场景，如企业自动化流程或研究实验。

EveryInc/compound-engineering-plugin

TypeScript · ★ 18,441 · 🍴 1,393 · 📈 349 stars today

Official Compound Engineering plugin for Claude Code, Codex, Cursor, and more

中文介绍 官方复合工程插件，适用于 Claude Code、Codex、Cursor 等开发工具。提供增强功能，帮助开发者更高效地处理复合型工程任务，如多文件协作、重构等。

affaan-m/ECC

JavaScript · ★ 199,378 · 🍴 30,615 · 📈 908 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 智能体性能优化系统，包含技能、本能、记忆、安全与以研究为先的开发模式。专为 Claude Code、Codex 等编码代理设计，提升其执行效率与稳定性，适合深度定制 AI 助手。

OpenBMB/VoxCPM

Python · ★ 22,803 · 🍴 2,673 · 📈 779 stars today

VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning

中文介绍 无 Tokenizer 的 TTS 模型，支持多语言语音生成、创意语音设计和高保真语音克隆。适合语音合成、虚拟助手、有声内容制作等场景，实现自然且富有表现力的语音输出。

galilai-group/stable-worldmodel

Python · ★ 1,472 · 🍴 166 · 📈 318 stars today

A platform for reproducible world model research and evaluation

中文介绍 用于可重复世界模型研究与评估的开源平台。提供标准化的数据集、训练框架和评价指标，帮助研究者统一对比不同世界模型的性能，适合强化学习与机器人领域。

Crosstalk-Solutions/project-nomad

TypeScript · ★ 27,377 · 🍴 2,685 · 📈 469 stars today

Project N.O.M.A.D, is a self-contained, offline survival computer packed with critical tools, knowledge, and AI to keep you informed and empowered—anytime, anywhere.

中文介绍 项目 N.O.M.A.D. 是一款自包含的离线生存计算机，集成关键工具、知识库和 AI，确保用户在无网络环境下依然获取信息并保持自主能力。适用于户外探险、应急场景或偏远地区。

run-llama/liteparse

Rust · ★ 7,935 · 🍴 467 · 📈 925 stars today

A fast, helpful, and open-source document parser

中文介绍 快速、开源且易用的文档解析库，可将 PDF、HTML 等文件提取为结构化数据。适用于 RAG 应用、知识库构建等场景，比传统解析器更轻量高效。

chen08209/FlClash

Dart · ★ 40,391 · 🍴 2,527 · 📈 187 stars today

A multi-platform proxy client based on ClashMeta,simple and easy to use, open-source and ad-free.

中文介绍 基于 ClashMeta 的多平台代理客户端，界面简洁易用、开源且无广告。支持 Windows、macOS 等系统，方便用户管理和切换代理规则，适合网络代理需求。

FareedKhan-dev/train-llm-from-scratch

Jupyter Notebook · ★ 2,307 · 🍴 375 · 📈 327 stars today

A straightforward method for training your LLM, from downloading data to generating text.

中文介绍 提供从头训练 LLM 的清晰流程，涵盖数据下载、预处理、训练到文本生成。适合想深入学习大模型训练原理的开发者或研究者，用于教学或实验。

ruvnet/RuView

Rust · ★ 68,961 · 🍴 9,197 · 📈 655 stars today

π RuView turns commodity WiFi signals into real-time spatial intelligence, vital sign monitoring, and presence detection — all without a single pixel of video.

中文介绍 通过 WiFi 信号实现空间智能、生命体征监测和存在检测，无需摄像头。利用商用 WiFi 设备的信号变化感知环境，适合隐私敏感场景如智能家居、老人看护。

DataTalksClub/data-engineering-zoomcamp

Jupyter Notebook · ★ 41,797 · 🍴 8,286 · 📈 274 stars today

Data Engineering Zoomcamp is a free 9-week course on building production-ready data pipelines. The next cohort starts in January 2026. Join the course here 👇🏼

中文介绍 免费的 9 周数据工程课程，教授构建生产级数据管道。涵盖数据摄取、转换、存储等主题，适合初学者或转行人士通过实践学习现代数据栈。下一期 2026年1月开班。

OpenMOSS/MOSS-TTS

Python · ★ 2,651 · 🍴 239 · 📈 62 stars today

MOSS‑TTS Family is an open‑source speech and sound generation model family from MOSI.AI and the OpenMOSS team. It is designed for high‑fidelity, high‑expressiveness, and complex real‑world scenarios, covering stable long‑form speech, multi‑speaker dialogue, voice/character design, environmental soun

中文介绍 MOSS‑TTS 系列开源语音和声音生成模型，由 MOSI.AI 与 OpenMOSS 团队推出。专注于高保真、高表现力，适用于复杂真实场景，如数字人配音、有声读物制作。

dreammis/social-auto-upload

Python · ★ 11,806 · 🍴 2,092 · 📈 73 stars today

自动化上传视频到社交媒体：抖音、小红书、视频号、tiktok、youtube、bilibili

中文介绍 自动化上传视频到抖音、小红书、视频号、YouTube、Bilibili 等主流社交平台。支持定时发布、批量操作，适合内容创作者和营销团队管理多平台分发。

anthropics/skills

Python · ★ 144,183 · 🍴 16,991 · 📈 454 stars today

Public repository for Agent Skills

中文介绍 Anthropic 官方 Agent Skills 公开仓库，提供一系列预定义技能供 Claude 等代理使用。开发者可借鉴或直接调用这些技能，快速增强 AI 助手的能力，适用于各种自动化场景。

codecrafters-io/build-your-own-x

Markdown · ★ 508,300 · 🍴 48,245 · 📈 817 stars today

Master programming by recreating your favorite technologies from scratch.

中文介绍 热门编程学习资源聚合，收录了从零构建各种技术（如数据库、编译器、Git）的教程。适合想深入理解底层原理的开发者，通过动手实践掌握核心技术。

Why Far Looks Up: Probing Spatial Representation in Vision-Language Models

👍 36

05/28 08:00

Vision-language models (VLMs) achieve strong performance on spatial reasoning benchmarks, yet it remains unclear whether this reflects structured 3D understanding or reliance on statistical shortcuts in natural images. We introduce a representation-level analysis framework that constructs minimal co

中文介绍 研究分析视觉语言模型（VLM）在空间推理中的表现，引入表征级分析框架，探讨其是否真正具备三维理解或仅依赖自然图像中的统计捷径。

DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation

👍 6

05/28 08:00

Robot manipulation critically depends on perception that preserves the action-relevant aspects of a scene. Yet most robot learning pipelines are built upon visual encoders pre-trained for static recognition or vision-language alignment, leaving motion understanding to downstream policies. We introdu

中文介绍 提出DynaFLIP框架，通过三模态动态引导表征改善机器人操作感知，使视觉编码器更好地理解运动信息，弥补静态识别预训练的不足。

Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection

👍 0

05/28 08:00

Recent advances in Vision-Language Models (VLMs) have achieved impressive performance across many tasks, yet prior studies report unsatisfactory performance when applying large language or multimodal models to finding abnormal patterns in sequential data. Public anomaly detection benchmarks typicall

中文介绍 提出一种高效的小型视觉语言模型用于时间序列异常检测，解决大型模型在序列数据异常模式识别中表现不佳的问题。

Reducing Political Manipulation with Consistency Training

👍 0

05/28 08:00

Large language models (LLMs) exhibit systematic political bias across a variety of sensitive contexts. We find that LLMs handle counterpart topics from opposing political sides asymmetrically. We refer to this phenomenon as covert political bias and identify 7 categories of techniques through which

中文介绍 研究发现大语言模型在敏感话题上存在系统性隐性政治偏见，提出基于一致性训练的方法减少此类操纵，识别出七种技术类别。

REPOT: Recoverable Program-of-Thought via Checkpoint Repair

👍 6

05/28 08:00

One-shot Program-of-Thought (PoT) emits a Python program that prints a primitive-action plan; a single invalid action silently invalidates the trajectory. We introduce RePoT (Recoverable PoT): a deterministic verified replay that walks the plan through the environment to its first invalid transition

中文介绍 提出REPOT（可恢复思维程序）方法，通过确定性验证重放机制，在程序化推理中遇到无效动作时进行修复，提升轨迹稳定性。

Xetrieval: Mechanistically Explaining Dense Retrieval

👍 16

05/28 08:00

Explaining why dense retrievers assign high relevance scores remains challenging because retrieval decisions are made through opaque high-dimensional embeddings. Existing explanations often focus on surface signals, such as lexical matches, token alignments, or post-hoc textual rationales, and thus

中文介绍 提出Xetrieval框架，从机制层面解释密集检索模型为何分配高相关性分数，克服现有方法仅依赖词汇匹配等表面信号的局限。

CoHyDE: Iterative Co-Training of LLM Rewriter & Dense Encoder for Tool Retrieval

👍 3

05/28 08:00

Tool retrieval over large API catalogs is a core bottleneck for LLM agents: user queries arrive in colloquial, often underspecified language, while the catalog uses technical API vocabulary that no fixed encoder can bridge on its own. The two dominant training approaches, contrastive encoder fine-tu

中文介绍 提出CoHyDE方法，通过迭代协同训练大语言模型改写器和密集编码器，解决工具检索中口语化查询与技术API术语间的语义鸿沟。

EarlyTom: Early Token Compression Completes Fast Video Understanding

👍 24

05/28 08:00

Video large language models (Video-LLMs) have demonstrated strong capabilities in video understanding tasks. However, their practical deployment is still hindered by the inefficiency introduced by processing massive amounts of visual tokens. Although recent approaches achieve extremely low token ret

中文介绍 提出EarlyTom方法，通过早期令牌压缩减少视觉令牌数量，加速视频理解任务，提升视频大语言模型的部署效率。

When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systems

👍 9

05/28 08:00

The design space of agentic AI inference spans two extremes: frontier large language models (LLMs), typically hosted in the cloud and offering strong performance across a wide range of tasks at substantially high cost, and more cost-efficient small language models (SLMs), which are amenable to on-de

中文介绍 研究云端大模型与设备端小模型在混合多智能体系统中的协同设计，权衡强性能与低成本之间的实际部署经验。

Thinking Before Constraining: A Unified Decoding Framework for Large Language Models

👍 5

05/28 08:00

Natural generation allows Large Language Models (LLMs) to produce free-form responses with rich reasoning, yet the lack of structure makes outputs difficult to verify. Conversely, constrained decoding ensures standardized formats but can inadvertently restrict reasoning capabilities by imposing cons

中文介绍 提出统一解码框架，结合自然生成与约束解码的优势，使大语言模型在保持推理能力的同时输出可验证的结构化结果。

PhyGenHOI: Physically-Aware 4D Generation of Dynamic Human-Object Interactions

👍 7

05/28 08:00

We address the task of generating physically accurate and visually faithful 4D Human-Object Interaction (HOI). Given a static 3D human and target object represented as 3D Gaussian Splats (3DGS), our goal is to synthesize dynamic scenes where the human actively engages with the object through actions

中文介绍 提出PhyGenHOI方法，从静态三维人体与物体（3D高斯泼溅表示）生成物理准确的动态人-物交互四维场景。

UniSteer: Text-Guided Flow Matching in Activation Space for Versatile LLM Steering

👍 20

05/28 08:00

Activation-based control steers large language models (LLMs) by intervening on their internal representations during inference, and has emerged as an effective paradigm for controlling behaviors such as persona and style. However, existing methods often rely on fixed steering directions or task-spec

中文介绍 提出UniSteer方法，在激活空间中使用文本引导的流匹配，实现灵活的大语言模型行为控制（如个性、风格）。

Discovering Cooperative Pipelines: Autoresearch for Sequential Social Dilemmas

👍 1

05/28 08:00

We study two-level autoresearch for cooperation: an outer-loop AI agent autonomously redesigns the inner-loop pipeline of an LLM policy-synthesis system for multi-agent Sequential Social Dilemmas (SSDs). A researcher agent R (run as a coding agent) reads the inner-loop source code, edits system prom

Verifiable Rewards Beyond Math and Code: Lightweight Corpus-Grounded Process Supervision for Factual Question Answering

👍 5

05/28 08:00

Applying reinforcement learning to improve factual accuracy in knowledge-intensive question answering faces a reward design dilemma. Response-level rewards provide only coarse supervision and cannot distinguish correct from incorrect statements within a reasoning trace. Sentence-level alternatives o

Towards Verifiable Multimodal Deep Research: A Multi-Agent Harness for Interleaved Report Generation

👍 8

05/28 08:00

Large Language Models (LLMs) have advanced autonomous agents from deep search, which retrieves concise factual answers, to deep research, which synthesizes scattered evidence into long-form reports. However, verifiable multimodal deep research remains challenging due to open-ended synthesis without

Colored Noise Diffusion Sampling

👍 16

05/28 08:00

Diffusion models achieve state-of-the-art image synthesis, with their generative trajectories fundamentally exhibiting a spectral bias, resolving low-frequency global structures early and high-frequency fine details later. Conventional stochastic differential equation (SDE) solvers fail to account f

CausaLab: A Scalable Environment for Interactive Causal Discovery Toward AI Scientists

👍 10

05/28 08:00

We introduce CausaLab, a scalable environment for evaluating interactive causal discovery by LLM agents. Unlike prior evaluations, CausaLab evaluates both whether an agent can solve a problem using causal evidence and whether its answer is grounded in a faithful recovered causal mechanism. Each epis

Token-Level Generalization in LoRA Adapter Backdoors: Attack Characterization and Behavioral Detection

👍 2

05/28 08:00

We show that LoRA adapters, the dominant distribution format for fine-tuned LLMs, can be reliably backdoored through training data poisoning while preserving baseline task performance. On a Qwen 2.5 1.5B prompt-injection classifier, a small fraction of poisoned examples drives a clean-accuracy-prese

YoCausal: How Far is Video Generation from World Model? A Causality Perspective

👍 38

05/28 08:00

As video diffusion models (VDMs) advance toward world models, a key question arises: do they truly understand causality, or merely overfit to statistical temporal patterns? Existing benchmarks mostly rely on synthetic data, limiting real-world generalization due to the sim-to-real gap. We present Yo

PhoneWorld: Scaling Phone-Use Agent Environments

👍 0

05/28 08:00

A central bottleneck for phone-use agents is that controllable, reproducible environments covering real mobile behavior are hard to build at scale. Existing mobile-agent benchmarks have made important progress on evaluation, but they do not by themselves provide a scalable way to construct many new

WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction

👍 5

05/28 08:00

Multimodal large language models are increasingly deployed as long-horizon agents, where memory must do more than recall: it must track an evolving world, revise what has gone stale, and surface the right evidence at decision time. Existing benchmarks measure recall over static dialogue, collapse me

PRISM: A Multi-Dimensional Benchmark for Evaluating LLM Peer Reviewers

👍 9

05/27 08:00

The rapid growth in submissions to machine learning venues has strained the scientific peer-review system and intensified interest in LLM-based automated peer reviewers. However, how good these systems are actually, especially compared to human reviewers at catching scientific gaps, remains poorly u

RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains

👍 7

05/27 08:00

Pointwise reward modeling offers critical signals for LLM post-training, yet struggles with absolute scoring in subjective, non-verifiable settings. Rubric-based methods address this by decomposing evaluation into explicit criteria, but existing approaches typically depend on frontier LLMs and suffe

Skill0.5: Joint Skill Internalization and Utilization for Out-of-Distribution Generalization in Agentic Reinforcement Learning

👍 18

05/27 08:00

Equipping large language models with explicit skills has emerged as a promising paradigm for enabling autonomous agents to solve complex tasks. Agent skills can be inherently divided into general skills for broad cognitive transfer and task-specific skills for dynamic execution. However, existing sk

PANDO: Efficient Multimodal AI Agents via Online Skill Distillation

👍 3

05/26 08:00

Recent advances in multimodal web agents often rely on increased inference-time computation, including rollout search, verifier passes, offline skill discovery, and specialist model stacks. This raises a central question: can a web agent become more efficient as it accumulates experience, rather tha

Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases

👍 2

05/26 08:00

Reinforcement Learning from Human Feedback (RLHF) is the standard method to align Large Language Models (LLMs) with human preferences. In this work, we introduce alignment tampering, a potential vulnerability where the LLM undergoing alignment influences the preference dataset, causing RLHF to ampli

CONF-KV: Confidence-Aware KV Cache Eviction with Mixed-Precision Storage for Long-Horizon LLM

👍 4

05/24 08:00

Long-horizon LLM inference turns the key--value (KV) cache into the dominant GPU memory consumer and makes per-token attention increasingly expensive. Many common eviction policies use static recency windows or historical attention, leaving unused a signal computed on every decoding step: the model'

Learning A Unified Risk Map for Autonomous Driving in Partially Observable Environments

👍 5

05/21 08:00

Occlusion-aware prediction remains a critical challenge in autonomous driving due to the inherent uncertainty of unobserved regions. Existing approaches either overestimate risk based on reachable states or struggle to predict accurate trajectories under high occlusion uncertainty. To address these

Reflective Prompt Tuning through Language Model Function-Calling

👍 3

05/20 08:00

Large language models (LLMs) have become increasingly capable of following instructions and complex reasoning, making prompting a flexible interface for adapting models without parameter updates. Yet prompt design remains labor-intensive and highly sensitive to formatting, phrasing, and instruction

ORACLE: Anticipating Scams from Partial Trajectories in Streaming App Usage

👍 0

05/09 08:00

Smartphone scams are increasingly prevalent and typically manifest as multi-stage, cross-application processes with gradually emerging intent. Effective intervention thus requires anticipating scams before the intent becomes explicit. This is inherently challenging, as decisions must rely on partial

How to Build a Software Factory with Claude Code That Ships Features While You Sleep

@sairahul1 · 106.0K 粉丝 · 2.8M 阅 · 1.5K 赞 · 204 转

05/25 16:46

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 分享一个 7 代理协作系统，让 Claude Code 实现从编码到功能交付的全自动化。指出大部分人只是用 AI 打字更快，而非真正发挥其潜力；这套系统能节省数月开发时间。

AI Agents: The Complete Course

@sairahul1 · 106.0K 粉丝 · 203.2K 阅 · 500 赞 · 82 转

05/24 16:25

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 系统梳理 AI 代理的完整指南，涵盖课程、书籍、实际构建和生产失败案例，帮助大家搞懂代理如何真正工作。强调很多人空谈概念但缺乏实操理解。

How to Build a Claude Research Agent That Reads the Internet Every Morning and Briefs You in 5 Mins

@cyrilXBT · 181.7K 粉丝 · 127.8K 阅 · 533 赞 · 80 转

05/25 11:16

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 教大家用 Claude 搭建一个晨间研究代理：自动阅读互联网，5 分钟内生成简报，取代花 20 分钟刷推文和邮件找重点的低效习惯。

How I Use Cursor

@poteto · 26.6K 粉丝 · 86.5K 阅 · 540 赞 · 48 转

05/26 02:15

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 个人分享使用 Cursor 的经验与反思。在 Meta 时因 Claude Code 火爆而付费使用，但面试前从未接触 Cursor；对比两款 AI 编码工具的实际感受。

Step-By-Step LLM Engineering Projects (2026 Edition)

@TheAhmadOsman · 59.9K 粉丝 · 54.5K 阅 · 512 赞 · 65 转

05/25 11:01

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 提供一套从零构建 LLM 的项目路线图：分词器→嵌入→位置编码→注意力→Transformer 块→解码→缓存→长上下文推理，强调动手实践优于理论阅读。

THE DIFF THAT CHANGED EVERYTHING

@difflawb · 20.3K 粉丝 · 21.9K 阅 · 1.1K 赞 · 389 转

05/25 23:13

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 Andrej Karpathy 在 2024 年发布的一段仅 40 行 shell 脚本，如何从个人实验演变为关键基础设施；探讨小改动带来的巨大影响。

how i make AI videos (a beginner’s breakdown)

@0xileri · 7.3K 粉丝 · 12.2K 阅 · 533 赞 · 63 转

05/25 02:10

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 以初学者角度拆分制作 AI 视频的流程，列出使用的工具和步骤，坦言还在学习中。分享过来人经验而非完备教程。

The Start of the End: AI Replacement Has Begun

@ActionModelAI · 57.1K 粉丝 · 5.8K 阅 · 505 赞 · 344 转

05/25 19:29

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 指出 AI 替代人类工作的经济变革已经开始，多数人仍未意识到其严重性。强调这不是科幻预言，而是正在发生的现实。

Browsers Are Dead. Codex Just Replaced Them.

05/28 22:56

中文介绍 认为传统浏览器已被取代，Codex 作为 AI 工具重新定义了用户与网页的交互方式。

AI Agents: Every NEW Claude & Codex Feature Explained

05/24 06:29

中文介绍 全面解析 Claude 和 Codex 的新功能，特别是 AI 代理方面的更新。

Codex: Build Your Full AI Marketing Team (Agents + Skills)

05/18 23:31

OpenAI just released Codex Mobile

05/16 02:58

Codex Mobile App Released (Complete Setup Guide)

05/15 09:35

Vibe Coding for Beginners (Full Course 2026)

05/08 03:28

Codex is The NEW Best AI Coding Tool (Here's Why)

05/03 04:35

7 Tools That Make Codex 10x MORE Powerful

04/30 09:19

Learn 95% of Codex in 30 minutes

04/29 10:08

With Codex and GPT 5.5 you can just do things.

04/26 07:48

[AINews] Founders and Forward Deployed Engineers

Latent Space · 05/30 09:57

a quiet day lets us highlight the new AIE WF focuses

中文介绍 本期 AINews 聚焦创使人及前沿部署工程师角色，并介绍 AIE WF 新关注点。

Boston Children’s uses AI to unlock new diagnoses

OpenAI News · 05/29 20:00

Boston Children’s Hospital uses OpenAI technology to improve patient care, reduce operational burden, and help diagnose more than 40 rare disease cases.

中文介绍 波士顿儿童医院利用 OpenAI 技术改善患者护理、减轻运营负担，并协助诊断超过 40 种罕见疾病。

How Braintrust turns customer requests into code with Codex

OpenAI News · 05/29 20:00

How Braintrust engineers use Codex with GPT-5.5 to run experiments and code faster.

中文介绍 Braintrust 工程师利用 Codex 与 GPT-5.5 加速实验和编码，将客户需求转化为代码。

How the Pope’s Magnifica Humanitas offers a template for individuals to meet the AI moment

MIT Tech Review AI · 05/29 18:00

Pope Leo XIV’s new encyclical on artificial intelligence includes a statement that warrants serious attention from technologists and policymakers: “Technology is never neutral.” Magnifica Humanitas (“Magnificent Humanity”) is a clarion call to all people to act with courage and solidarity as we ente

中文介绍 教皇 Leo XIV 发布新通谕《辉煌人性》，强调「技术从不中立」，呼吁个人勇敢应对 AI 时代。

not much happened today

Smol AI News · 05/29 13:44

**Anthropic** rolled out **Claude Opus 4.8**, which shows incremental improvements but mixed benchmark results, including better cooperation and coding behavior but some regressions in document parsing. Platform updates include mid-conversation system instructions enhancing long agent sessions, thou

中文介绍 Anthropic 发布 Claude Opus 4.8，改进协作与编码行为，但文档解析出现退步，平台新增对话中系统指令功能。

Strengthening societal resilience with Rosalind Biodefense

OpenAI News · 05/29 11:00

OpenAI launches Rosalind Biodefense, expanding trusted access to GPT-Rosalind for vetted developers and U.S. government partners advancing biodefense, public health, and pandemic preparedness through frontier AI.

中文介绍 OpenAI 推出 Rosalind Biodefense，为受信任开发者和美国政府合作伙伴提供 GPT-Rosalind，助力生物防御与公共卫生。

[AINews] Anthropic raises $965B Series H, releases Opus 4.8 and Dynamic Workflows/ultracode

Latent Space · 05/29 10:07

Total Anthropic victory!

中文介绍 Anthropic 以 9650亿美元估值完成 H 轮融资，发布 Opus 4.8 及动态工作流/超代码功能。

A shared playbook for trustworthy third party evaluations

OpenAI News · 05/29 08:00

OpenAI shares guidance on third-party AI evaluations, covering how to assess model capabilities, safeguards, and validity for frontier systems.

中文介绍 OpenAI 发布第三方评估指南，涵盖模型能力、安全防护及有效性评估方法，旨在建立可信评估框架。

Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

Hugging Face Blog · 05/29 08:00

中文介绍 本文是 PyTorch 性能分析系列首篇，介绍 torch.profiler 的基础使用方法。

Opus 4.8 🧠, Anthropic at $965B 💰, Microsoft’s coding model 👨‍💻

TLDR AI · 05/29 08:00

中文介绍 今日 AI 摘要：Opus 4.8 发布、Anthropic 估值达 9650亿美元、微软推出新编码模型。

The Age of Async Agents — Cognition's Walden Yan & OpenInspect's Cole Murray

Latent Space · 05/29 02:41

80% Devin Commits, Spec-to-PR Workflows, Full VMs, Agent Memory, and PMs Shipping Code

中文介绍 Cognition 的 Walden Yan 与 OpenInspect 的 Cole Murray 讨论异步代理，涵盖 80% Devin 提交、Spec-to-PR 工作流及智能体记忆。

How Endava builds an agentic organization with Codex

OpenAI News · 05/28 20:00

Learn how Endava uses Codex to build an agentic organization, accelerating software delivery and reducing requirements analysis from weeks to hours.

中文介绍 Endava 利用 Codex 构建智能体型组织，将需求分析时间从数周缩短至数小时，加速软件交付。

The AI Hype Index: AI gets booed in graduation season

MIT Tech Review AI · 05/28 17:51

It is one thing to say AI will change the world. It is another to expect the class of 2026 to applaud it. In fact, when former Google CEO Eric Schmidt told University of Arizona graduates that their task is to help shape AI, he was met with a resounding chorus of boos. “I can…

中文介绍 毕业季 AI 遭嘘：前谷歌 CEO Eric Schmidt 在亚利桑那大学演讲时，呼吁毕业生塑造 AI 却引来一片嘘声。

[AINews] Cognition raises $1B in $26B Series D

Latent Space · 05/28 15:26

coding is an uncapped TAM market

中文介绍 Cognition 以 260亿美元估值完成 D 轮融资 10亿美元，强调编程市场总规模无上限。

Anthropic raises $65B in Series H at a $965B post-money valuation, releases Opus 4.8 and Dynamic Workflows

Smol AI News · 05/28 13:44

**Anthropic** announced a massive **$65B Series H financing** at a **$965B valuation**, led by **Altimeter, Dragoneer, Greenoaks, and Sequoia**, with run-rate revenue surpassing **$47B**. They launched **Claude Opus 4.8**, an update to Opus 4.7 featuring "sharper judgment," "more honesty," and longe

中文介绍 Anthropic 完成 650亿美元 H 轮融资，估值达 9650亿美元，年化收入超 470亿美元，同时发布 Claude Opus 4.8 及动态工作流。

AI 编程时代， MVP 思维已经失效了？

05/30 19:25

21 回复 · 程序员 节点

以防大家不知道怎么让 Agent 一直干活不停下来

05/30 18:05

8 回复 · 程序员 节点

大模型不听话咋整（skill）

05/30 17:29

4 回复 · 程序员 节点

Claude Code 团队计划，非中转站

05/30 17:18

11 回复 · 程序员 节点

大佬们 vibe coding 的项目怎么样了

05/30 17:17

54 回复 · 程序员 节点

Kimi Code 用起来感觉还行，就是很容易达到限额

05/30 12:31

25 回复 · 程序员 节点

关于 iPhone air 国行 esim 卡的一个想法

05/30 10:49

9 回复 · Apple 节点

美区礼品卡被连着砍单 3 次

05/29 17:40

17 回复 · Apple 节点

计划外，换机请教🙏

05/29 11:18

6 回复 · Apple 节点

求助｜ iOS App Store 提审长时间卡审核，状态无更新

05/29 11:14

6 回复 · Apple 节点

Microsoft degrades functionality of perpetually-licensed offline products

05/31 07:26

437 points · 139 comments

Cheese Paper: a text editor specifically designed for writing

05/31 06:58

59 points · 9 comments

Shantell Sans

05/31 06:06

97 points · 10 comments

The AV2 Video Standard Has Released (Final v1.0 Specification)

05/31 05:46

32 points · 0 comments

Rotary GPU: Exploring Local Execution for Large MoE Models Under Limited VRAM

05/31 05:05

23 points · 4 comments

Design Engineering Magazine

05/31 04:46

63 points · 6 comments

wolfSSL releases a new product; wolfCOSE a zero alloc C embbedded COSE stack

05/31 04:42

69 points · 13 comments

Domain expertise has always been the real moat

05/31 04:40

315 points · 192 comments

Dusklight – GC Twilight Princess Decompiled

05/31 04:24

72 points · 8 comments

Parallel Reconstruction of Lawful TLS Wiretapping

05/31 03:47

61 points · 31 comments

Jef Raskin, the Visionary Behind the Mac (2013)

05/31 03:42

76 points · 37 comments

Show HN: 500 years of Joseon court omens as an observability dashboard

05/31 03:23

80 points · 13 comments

Zig ELF Linker Improvements Devlog

05/31 01:29

176 points · 52 comments

OpenRouter raises $113M Series B

05/31 01:27

366 points · 181 comments

Microcode inside the Intel 8087 floating-point chip: register exchange

05/31 01:27

90 points · 16 comments

Accenture to acquire Ookla

05/31 00:28

https://www.theverge.com/tech/889234/downdetector-ookla-spee..., https://archive.ph/FR8NDhttps://arstechnica.com/information-technology/2026/03/downd...

Werner Herzog in conversation with Paul Cronin (2014)

05/31 00:13

75 points · 24 comments

Voxel Space (2017)

05/30 22:25

253 points · 56 comments

Openrsync: An implementation of rsync, by the OpenBSD team

05/30 18:51

325 points · 143 comments

Leo's first encyclical attacks technological messianism

05/30 18:30

191 points · 210 comments

Pandoc Templates

05/30 17:56

366 points · 48 comments

Zig: Build System Reworked

05/30 16:38

321 points · 210 comments

I found a seashell in the middle of the desert

05/29 10:49

232 points · 64 comments

Gustav Klimt and Egon Schiele in Conversation (2018)

05/29 06:29

19 points · 3 comments

Show HN: Open Envelope – an open schema for defining AI agent teams

05/29 04:30

Built an open JSON Schema for defining AI agent teams.Multi-agent systems are becoming a real deployment pattern — not single assistants, but teams with roles, handoffs, and human checkpoints. But there's no shared way to define one that travels across frameworks. Every implementation is scatte

Racket v9.2 is now available

05/28 17:42

30 points · 1 comments

C++ CLI for folder encryption with AES-256-GCM and USB-based key loading

05/28 12:06

I built a Linux CLI tool that encrypts and decrypts folders using AES-256-GCM. It also hides file and folder names and stores the mapping in an encrypted file.Repo: https://github.com/sahilPadmani/ACE-files-encryption

90% of the T Distribution

05/27 21:29

29 points · 7 comments

Navier-Stokes fluid simulation explained with Godot game engine

05/27 06:15

189 points · 24 comments

It takes two neurons to ride a bicycle (2004)

05/27 00:51

98 points · 46 comments