1
对标 Reddit，Meta 悄然推出“Forum”应用
Meta推出对标Reddit的“Forum”独立群组应用，聚焦深度交流与兴趣社群。
IT之家 5 月 24 日消息，Meta 公司悄然推出一款名为“Forum（论坛）”的独立群组应用。该产品定位对标 Reddit，官方将其描述为“专为深度交流、真实答疑和兴趣社群打造的专属空间”。 用户登录 Facebook 账号后，应用会同步个人群组、资料与动态，也可以使用昵称发布帖子，操作方式和…
2
Meta quietly launches a new Reddit-like app called Forum
Meta低调上线类Reddit应用Forum，或开启社交新战场。
The company describes the app as a "dedicated space built for deeper discussions, real answers and communities you care about."
3
Ask HN: Forbid Reddit HN Submissions?
Reddit充斥AI垃圾内容，HN社区热议是否该封杀其提交，引发对内容治理的深层思考。
cc @dang Reddit has been filling up with AI generated content. Either bots engaging with naïf audiences who don't realize they're bots, or supposed hu…
4
Hiding in Plain Sight: Finding MAHA on Reddit
Reddit上隐藏着一个叫MAHA的神秘社区？这篇论文用数据挖掘方法把它揪了出来，揭秘数字暗流。
arXiv:2605.20435v1 Announce Type: cross Abstract: Make America Healthy Again (MAHA) is a national health movement that encompasses a striking mix of b…
5
A new EDIT tool for LLM agents
Redis作者分享LLM agents中EDIT工具的替代方案，实用技巧提升代码生成效率。
Article URL: https://antirez.com/news/166 Comments URL: https://news.ycombinator.com/item?id=48190359 Points: 5 # Comments: 1
6
Fully Open Meditron: An Auditable Pipeline for Clinical LLMs
完全开源且可审计的临床大模型流水线，解决AI医疗黑箱问题，数据来源与训练过程全透明。
arXiv:2605.16215v1 Announce Type: new Abstract: Clinical decision support systems (CDSS) require scrutable, auditable pipelines that enable rigorous, …
7
Registers Matter for Pixel-Space Diffusion Transformers
一键直达最新扩散变换器论文，深入理解寄存器标记在像素空间训练中的关键作用，紧跟前沿研究
arXiv:2605.16147v1 Announce Type: new Abstract: Vision Transformers (ViTs) are known to exhibit high-norm patch-token outliers that degrade feature ma…
8
AI Faceless Video Generator for TikTok, Shorts and Reels
一个让Reddit故事创作者效率飙升10倍的AI视频生成器：从一天一视频到10分钟搞定，专为TikTok、Shorts和Reels设计，无脸出镜也能轻松吸粉。
Article URL: https://faceless-video.com/ Comments URL: https://news.ycombinator.com/item?id=48165742 Points: 1 # Comments: 0
9
AI Memory Reader – Native macOS app for browsing Claude Code memory files
一款原生macOS/iOS应用，专门用于浏览AI代理（如Claude Code）的记忆文件。支持Markdown渲染、编辑、搜索和AI工具集成，让管理AI记忆变得高效直观。
Article URL: https://github.com/nvwalj/ai-memory-reader Comments URL: https://news.ycombinator.com/item?id=48164406 Points: 3 # Comments: 0
10
Stop letting npm install run untrusted code on your machine — meet np-audit
每次`npm install`都可能为恶意代码敞开大门。Shai-Hulud蠕虫已感染数百个包，窃取数万机密。`np-audit`工具在npm执行前静态分析安装脚本，阻止预安装攻击。立即采取行动保护你的开发环境。
npm install You type it dozens of times a day. You probably typed it this morning. And every time you did, you handed arbitrary code execution to ever…
11
Show HN: Agenda Intel MD – schemas and CLI to audit LLM strategic-risk briefs
一个为战略情报AI代理打造的审计层：用JSON schema、CLI和MCP协议，让LLM输出从“听起来合理”变成“可审计可追溯”——开源、可直接CI集成，适合政策、制裁、地缘风险等需要严格证据链的场景。
Article URL: https://github.com/vassiliylakhonin/agenda-intelligence-md Comments URL: https://news.ycombinator.com/item?id=48160912 Points: 1 # Commen…
12
Ask HN: What happened to ssh-audit.com?
ssh-audit.com 的 DNS 出事了：根服务器残留胶水记录，但两个 NS 一个失联，一个返回 SERVFAIL。如果你了解内情，或认识站主，赶快去 HN 支个招。
The are still glue records in the root servers and the name is still registered but one name server no longer responds and the other gives SERVFAIL. I…
13
Predictive Maps of Multi-Agent Reasoning: A Successor-Representation Spectrum for LLM Communication Topologies
别等跑完实验才后悔选错通信拓扑——这篇工作用后继表示谱量（谱半径、谱隙、条件数）提前诊断链、星、网格等拓扑的漂移、共识和鲁棒性，且条件数完美排序扰动鲁棒性（rs=1.0），把多智能体LLM系统设计从试错变成数学可预测。
arXiv:2605.11453v2 Announce Type: replace-cross Abstract: Practitioners deploying multi-agent large language model (LLM) systems must currently choose…