1
微软 Win11 原生 WinUI 框架现已开源
IT之家 8 月 29 日消息，据科技媒体 neowin 今天报道，微软现已完成 WinUI 开源工作，该项目的主线开发工作已在 GitHub 平台公开进行。作为参考，WinUI 是 Windows 11 操作系统使用的原生 UI 框架。 IT之家了解到，微软于一年多以前宣布开源 WinUI。如今微…
2
Anthropic 安全框架自动降级：Claude 误删开发者 700 GB 主目录，自动化清理反成灾难
IT之家 8 月 29 日消息，开发者 Sebastien Guillemot 本周三在 X 上反馈称，他在测试 AI 智能体的文件删除防护机制时遭遇严重事故，Claude 直接给他删除了大约 700GB 数据，包括整个用户主目录，相当于一周的工作成果。 据称，事故发生前，Anthropic 的安全…
3
RASET: Router-Agnostic Safety-Critical Expert Tuning Exposes Localized Safety Enforcement Failures in Mixture-of-Experts LLMs
MoE大模型的安全防线可能只藏在少数专家里，RASET框架专攻这一漏洞并揭示其风险。
arXiv:2605.29708v2 Announce Type: replace Abstract: Mixture-of-Experts (MoE) LLMs rely on sparse, router-driven expert activation, yet how safety alig…
4
Show HN: Pyxle – Python and React in one file, no API layer
Pyxle 把 Python 与 React 塞进同一文件并省掉 API 层，SSR 速度比 Next.js 快约 2.2 倍，全栈开发从未如此清爽。
Article URL: https://pyxle.dev Comments URL: https://news.ycombinator.com/item?id=49433709 Points: 1 # Comments: 0
5
Beyond RGB: Benchmarking and Enhancing MLLMs for Hyperspectral Image Understanding via Training-Free Reasoning Framework
多模态大模型不止看懂RGB，高光谱成像理解迎来全新基准与免训练推理框架。
arXiv:2604.08884v2 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) have achieved strong performance on RGB image under…
6
Show HN: Crilio – pytest for prompts
用pytest风格测试LLM提示词，Crilio让AI输出验证变得可编程。
Article URL: https://github.com/mukundzha/crilio Comments URL: https://news.ycombinator.com/item?id=49433864 Points: 5 # Comments: 0
7
AdaR: A Framework for Equipping LLMs with Adaptive Reasoning
让大模型按需调整推理深度，AdaR框架为复杂问题动态分配计算资源，兼顾效率与准确。
arXiv:2510.04617v3 Announce Type: replace Abstract: Mathematical reasoning is a primary indicator of large language models (LLMs) intelligence. Howeve…
8
RAG vs. Fine-Tuning: The AI Engineer's Decision Framework
RAG与微调如何选？一文给出AI工程师的决策框架，帮你告别盲目调参，精准落地生成式AI应用。
The Senior Engineer’s Decision Framework: RAG vs. Fine-Tuning For many engineering teams, the journey into Large Language Models (LLMs) often begins w…
9
When Do LLMs Replace Fine-Tuned NLU? A Decision Framework for Intent Detection in Production Conversational Systems
零样本大模型能否替代精调NLU？这项研究用ATIS和CLINC150实测，给出基于意图空间的决策框架，生产环境选型必看。
arXiv:2608.20371v1 Announce Type: cross Abstract: A common claim is that zero-shot large language models (LLMs) can replace fine-tuned NLU classifiers…
10
AEGIS: Preventing Cross-Domain Resource Abuse in MCP
首个针对MCP跨域资源滥用防护框架登场，为AI协议安全加装“门禁系统”
arXiv:2608.20481v1 Announce Type: cross Abstract: The Model Context Protocol (MCP) is an open source JSON-RPC protocol that standardizes how large lan…
11
Are you good at AI, or just using it?
用五级阶梯自测AI熟练度，从新手到高手，清晰定位你的AI能力水平。
We’re working on a ladder for individual AI proficiency and would love feedback on both the levels and the definitions. L0 New: brand new to AI, or ha…
12
Large Language Models: A Mathematical Formulation
从数学底层重新定义大语言模型，为理解LLM提供严谨的理论框架。
arXiv:2601.22170v2 Announce Type: replace-cross Abstract: Large language models (LLMs) process and predict sequences containing text to answer questio…
13
A shot-scraper-style JSON API on Bun 1.4's new Bun.WebView
Bun 1.4 新 WebView 让你轻松构建页面渲染 + JSON 抽取的轻量 API，替代 Shot-scraper 更简单高效。
Research: A shot-scraper-style JSON API on Bun 1.4's new Bun.WebView Today saw the long awaited release of Bun 1.4 , the first stable version sin…
14
Eval4Sim: An Evaluation Framework for Persona Simulation
首个面向LLM人格模拟的评估框架，专测虚拟人设是否忠实还原真实人类行为。
arXiv:2603.02876v2 Announce Type: replace Abstract: Large Language Model personas, explicit profiles specifying a user's attributes, preferences, and …
15
TSQueryBench: LLM-as-a-Judge for Time Series Explanations
用LLM当裁判，给时间序列解释打分，评估框架新思路值得关注。
arXiv:2604.02118v2 Announce Type: replace Abstract: Natural language explanations of time series data are increasingly produced by foundation models i…
16
DeepSeek Harness vs. LangChain Deep Agents – what's the actual difference?
AI 智能体框架之争：DeepSeek Harness 与 LangChain Deep Agents 的差异到底在哪？快速定位技术选型关键点
What is DeepSeek Harness and why is it different from Deep Agents? Comments URL: https://news.ycombinator.com/item?id=49345619 Points: 2 # Comments: 0
17
The Value of a Prompt: An LLM-Relative Kolmogorov-Complexity Approach
用列文-柯尔莫哥洛夫复杂度度量提示词价值，给“好提示”一个可计算的理论标尺。
arXiv:2608.16438v1 Announce Type: new Abstract: In a world where valuable artifacts are increasingly created, completed, or processed by LLMs, the cen…
18
Mental Model Management: An Operator-Based Framework for LLM Memory
用算子式框架管理LLM心智模型，破解长文本记忆压缩与动态更新的难题。
arXiv:2608.15451v1 Announce Type: new Abstract: Large language models process large amounts of information but usually lack an explicit mechanism for …
19
Pairton: Iterative Reconstruction of Short-Lived Particles
用图结构做掩码预测，迭代重建短寿命粒子，物理分析新思路。
arXiv:2608.14278v1 Announce Type: cross Abstract: We present Pairton, an iterative framework for reconstructing short-lived particles in high-energy c…
20
微信小程序组件框架升级：glass-easel 带来更多特性及更优性能
升级小程序组件开发体验，glass-easel 框架提供更强特性与 WebView 渲染支持，性能更优。
IT之家 8 月 15 日消息，微信开发者本周（8 月 12 日）宣布，微信小程序新一代框架组件 glass-easel 已经可以自由使用，该框架具备更多特性和更优性能， 适合用来编写、维护复杂的小程序 。 据悉，glass-easel 是新版微信小程序组件框架的核心实现。从本质上说，glass-e…