2026年11月4日 技术热点总结

项目简介：手机隐私审计：用 adb 扫出 Android 应用的权限、上报端点与数据流向，并给出可回滚的处置。Desktop Android leak audit: permissions, endpoints, data flows, reversible fixes.

技术栈：Kotlin、adb、android、android-privacy、android-security、android-tools、apk-analysis、bloatware、cli

核心介绍：NOIR 是一个跑在电脑上的审计工具，通过 adb 读取已连接 Android 设备的应用清单、权限和 APK 里的网络端点，给出分类、风险评分和可执行的处置建议。它也可以在需要时调用 jadx 对单个应用做反编译和关键字检索。 > 定位：审计与自我防护。面向自己的设备、自己的应用、已授权的安全研究。 1. 盘点：列出所有「有网络权限 + 敏感权限」的应用，以及它们各自拿到了什么。 2. 抽取端点：在设备端用 unzip + strings 从每个 APK 的 dex 里提取 URL，只回传几 KB，不把整包拉回电脑。 3. 分类与打分：按包名和域名识别厂商（Xiaomi / Google / 腾讯 / 高德 …）与类别（遥测、广告、定位、输入法 …

项目简介：A Claude Code skill that turns any machine into a 3D simulator you can take apart

技术栈：JavaScript

核心介绍：A Claude Code skill that turns any machine into a 3D simulator you can take apart. Ask how something works. Claude builds it as one offline HTML page: the machine runs in the middle, every panel around it reads the real numbers, a button opens the part doing the work, and when nobody touches it the page tours itself.

项目数据：⭐ 1,641 Stars，🍴 251 Forks

项目简介：Agent skill: describe your stack in a sentence, get an architecture diagram with official logos

技术栈：TypeScript、agent-skills、architecture-diagram、claude-code、claude-skills、codex、diagram、system-architecture、tech-stack

核心介绍：Describe your stack in a sentence. Get an architecture diagram with official logos. > Diagram our stack: Next.js and a Hono API on Vercel. Postgres on Neon, auth with Clerk, payments with Stripe. GitHub Actions deploys. npx skills add caru-ini/stack-diagram Works with Claude Code, Codex, Cursor and other agents that support skills. In the Claude apps, u…

Few-step autoregressive video diffusion enables efficient generation, yet maintaining consistency over long rollouts remains challenging, especially when subjects or scenes reappear after going out of sight. Despite progress in fidelity and rollout robustness, existing distillation methods do not fully exploit richer historical information to constrain the student's current prediction. To address this gap, we introduce History Forcing, a unified on-policy distillation framework that strengthens historical supervision for few-step autoregressive video generators. The framework uses the resid…

Research taste, the judgment of what to read, which problems to pursue, and which methods to try, increasingly shapes the work of AI scientists, yet it remains far less understood than their ability to execute research. We introduce IdeaTaste, a benchmark that scores an LLM by how closely the research proposals it writes from a literature snapshot frozen at 2025 match the work the same communities published in 2026. Our experiments span six contemporary language models and four research areas from established to emerging fields, covering more than 90{,}000 proposals in total. LLM agents fal…

On-policy distillation (OPD) trains a student on its own generations by matching the teacher's token probabilities. These probabilities entangle two kinds of information: which tokens the teacher prefers, and how confident it is. The latter reflects the teacher's own capacity and calibration, which a smaller student may be unable to reproduce. Because the teacher temperature changes these values without changing the ordering, OPD is also sensitive to the temperature choice. We ask whether the teacher's ordering of candidate tokens is enough, and propose Order-Only On-Policy Distillation (O^…

Low-rank SVD-based compression offers a powerful strategy to reduce the computational costs of LLMs. However, existing methods face two key limitations: (i) global rank allocation, where uncalibrated error proxies fail to capture complex error propagation, and (ii) decomposition quality, where Fisher-based estimators suffer from severe rank collapse. In this work, we address these limitations by introducing Layer-wise Error Modeling Search (LEMS) and KFAC-SVD. LEMS advances rank allocation by introducing a layer-wise error surrogate that integrates local and global layer importance alongsid…

Linear attention scales linearly with sequence length, but its finite-state memory can limit long-context retrieval. In gated linear recurrences, memory retention depends on state decay, which is shaped jointly by gate parameters A_log and dt_bias and by input activations. In controlled Gated DeltaNet (GDN) experiments, both parameters remain highly correlated with their initial values after training, highlighting the importance of initialization. Therefore, we introduce Dual-Axis Memory Initialization (DAMI), which jointly configures these parameters before training. DAMI assigns geometric…

Complex documents such as scientific papers contain rich information in tables, figures, and equations; however, existing deep research (DR) agents typically reduce papers to OCR-extracted text. We instead ask whether vision-language models (VLMs) can enable page-native deep research: directly retrieving and reasoning over rendered document page images. We introduce DR Pixel, the first open-source end-to-end multimodal DR agent that uses rendered page images as its sole retrieval and reasoning unit, operating over a corpus of 1.06M pages. To evaluate this setting, we construct DRPixelBench,…