从桌面美化到AI同事：今日开源项目里的实用主义与实验精神

收录数按本期首次达标收录的项目统计。

• SplendorHornbill/windows-animated-wallpapersPython · 300 Stars

  动态壁纸

  Stars 统计更新于 2026/11/4 07:46:41

  查看原始资料

  作者描述

  High-quality animated and live wallpapers for Windows 10 and 11. Smooth 60 FPS animations, low CPU usage, 4K support. Open-source and portable

  编报时的 AI 中文解读

  为Windows 10/11提供高质量动态壁纸的合集，支持4K、60FPS动画，低CPU占用，可自定义设置，适合追求桌面美观的用户。

  编报时的 README 依据

  > Works seamlessly with Windows 10/11. Supports 1080p, 1440p, and 4K resolutions. --- ## 🚀 Quick Start <p align="center"> <a href="https://github.com/YOUR_USERNAME/windows-animated-wallpapers/archive/refs/heads/main.zip"> <img src="https://img.shields.io/badge/Download%20Project-2ea44f?style=for-the-badge&logo=download&logoColor=white" alt="Download Project"> </a> </p> 1. Download the project using the button above. 2. Extract the archive into a separate folder. 3. Run **WallpaperSetup.exe**. 4. Follow the on-screen instructions.

  | 🎨 **Customizable** | Adjust speed, opacity, and sound reactivity | | 📦 **Portable** | No installation required — just run the `.exe` | --- ## 🧩 How to use 1. Download and extract the archive. 2. Run **WallpaperSetup.exe**. 3. Browse the `wallpapers/` folder to preview files. 4. Select a wallpaper and click "Set as Desktop Background". 5. Right-click the tray icon to adjust performance settings. --- ## ⚙️ Configuration All settings are available in the system tray menu: - **Performance Mode**: Balanced / High Quality / Battery Saver.

  - **Pause on Fullscreen**: Automatically stops animation when gaming. - **Startup**: Launch with Windows. - **Audio Reactivity**: Sync animations with system audio (optional). --- ## ❓ FAQ **Will this slow down my PC?** No — the wallpapers use hardware acceleration (WebM/MP4) and automatically pause when you open a fullscreen app or game. **Does it work on Windows 10?** Yes, it supports both Windows 10 (version 1903+) and Windows 11. **Can I add my own videos?** Yes — just drop `.mp4` or `.webm` files into the `wallpapers/` folder. **Is it safe?**

  查看 README 缓存打开 GitHub
• AhmadHassan-BTed/rain_generatorPython · 1,587 Stars

  雨声模拟、音频导出

  Stars 统计更新于 2026/11/4 07:51:14

  查看原始资料

  作者描述

  This repository will be used to develop a rain generator, actually it use white noises to simulate wind and have an integrated thunderstorm simulator.

  编报时的 AI 中文解读

  这是一个用Python和Pygame构建的程序化雨和雷暴模拟器，提供多图层视差降雨、粒子飞溅、分形闪电和实时程序化音频合成，适合用于放松、环境音生成或视觉演示。

  编报时的 README 依据

  ## Features - **Parallax Rainfall:** Multi-depth rain layers with independent velocities, lengths, opacities, and wind deflection angles. - **Particle Splashes & Ripples:** Water droplets eject upward on surface collision with expanding ground ripples. - **Fractal Branching Lightning:** Midpoint displacement generation producing branching lightning bolts, multi-phase sky illumination, and camera trauma shake. - **Procedural Sound Synthesis:** - Real-time lowpass filtered rain ambiance. - Granular droplet impact synthesis using decaying sinusoids.

  - `-o`, `--export`: Output filepath to generate and export audio directly without GUI ### Audio Track Export Generate an ambient audio file of any custom length (e.g. 2 minutes of thunderstorm audio): ```bash python thunderstorm_simulator.py --duration 120 --export thunderstorm_2m.wav --preset thunderstorm ``` You can also click the **Export Audio** button in the top HUD (or press `G`) while running the simulator to open the interactive generator dialog. --- ## Controls & Keybindings | Key | Action | | --- | --- | | `F` / `F11` | Toggle Fullscreen mode |

  ```bash git clone https://github.com/azuletto/rain_generator.git cd rain_generator ``` 2. Install dependencies: ```bash pip install -r requirements.txt ``` 3. (Optional) Install in editable mode: ```bash pip install -e . ``` --- ## Usage Run via the main launcher: ```bash python thunderstorm_simulator.py ``` Or via module execution: ```bash python -m rain_generator ``` ### CLI Arguments ```bash python -m rain_generator --preset thunderstorm --fullscreen python -m rain_generator --preset drizzle --mute -W 1280 -H 720 ``` Available options:

  - `-p`, `--preset`: Weather preset (`drizzle`, `steady`, `thunderstorm`, `gale`, `cozy`) - `-i`, `--intensity`: Intensity level (`Low`, `Medium`, `High`, `Extreme`) - `-f`, `--fullscreen`: Launch in fullscreen mode - `-m`, `--mute`: Launch with audio muted - `-W`, `--width`: Window width in pixels (default: 1080) - `-H`, `--height`: Window height in pixels (default: 720) - `--fps`: Target framerate (default: 60) - `--no-shake`: Disable camera shake on lightning - `-d`, `--duration`: Audio export duration in seconds (default: 60)

  # GenRainpy - Rain and Thunderstorm Simulator [](https://www.python.org/) [](https://www.pygame.org/) [](LICENSE) A procedural rain and thunderstorm simulator built with Python, Pygame, NumPy, and SciPy. Features multi-layer parallax rainfall, particle splash physics, fractal branching lightning strikes, and real-time procedural audio synthesis. ---

  查看 README 缓存打开 GitHub
• RikiyaT/lm_understand-logsPython · 726 Stars

  实验日志

  Stars 统计更新于 2026/11/4 07:46:35

  查看原始资料

  作者描述

  Complete logs of the experiments in the blog post 'LLMs can understand and write neural network weights'

  编报时的 AI 中文解读

  提供关于LLM理解和编写神经网络权重实验的完整日志，包含提示、回复、代码及评估，适用于研究人员复现或分析实验。

  编报时的 README 依据

  # Logs: LLMs can understand and write neural network weights The complete logs of every experiment in the blog post *Sparks of Weight Space Programming: LLMs can understand and write neural network weights*. For each experiment you get: - everything we sent the model; - everything it sent back, including its reasoning summaries; - every piece of code it ran, with the output it saw; - the evaluation behind each number in the post; - the networks it read or wrote. Each experiment folder has its own README. It says which run is which tab or row in the post, and which file each

  number comes from. | Folder | In the post | Runs | Tools | |---|---|---|---| | [`exp1_which_game/`](exp1_which_game) | Experiment 1: which game was this network trained on? | Breakout, Pong, Pac-Man, Freeway (open answer) | none | | [`exp2_diffusion/`](exp2_diffusion) | Experiment 2: what does this diffusion model generate? | circle, square, triangle | none, no forward pass | | [`exp3_language/`](exp3_language) | Experiment 3: which language a language model learned, from its middle blocks | Python, romaji, English, two conversations each | Python sandbox, no forward pass |

  | [`exp8_math/`](exp8_math) | Experiment 8: make a language model better at math | Qwen3.5-0.8B with 10, 25 and 50 passes, and the Gemma 3 270M follow-up | Python sandbox | Only the runs the post reports are here. Trial runs of earlier prompt versions are left out. ## How the model was run "Astra" in the post is OpenAI's `gpt-6-astra`. **One turn.** Every turn is one call of the Codex CLI: `codex exec`, or `codex exec resume <thread>` to continue a conversation. The prompt goes in on stdin. The reasoning effort is set per call and recorded in each turn's `run.json`:

  of one conversation still match up. ## Not included - **The Gemma 3 270M fine-tunes of Experiment 7** (17 GB). The data, the training script and the hashes are there, so they can be rebuilt. - **The intermediate files the model wrote while working:** activations, generated-text dumps and its draft networks. The final networks and weight edits are included, Experiment 8's as exact diffs against the public base weights, with a script that applies them. - **The text of Experiment 3's Python corpus.** It is other projects' code, so we give a manifest and a rebuild script instead.

  - **The GPQA question texts in Experiment 8's generations.** GPQA's scores and per-question results are kept. ## Licenses See [`LICENSE`](LICENSE). | Content | License | |---|---| | Logs, data and networks | CC BY 4.0 | | Our scripts | MIT | | MinAtar source in Experiment 5's prompts | GPL-3.0, see `exp5_write_from_scratch/LICENSE-MinAtar` | Third-party benchmark data keep their own licenses: MATH, GSM8K, IFEval, MMLU-Pro and the others in Experiment 8.

  查看 README 缓存打开 GitHub
• rerun-build/rerunTypeScript · 575 Stars

  AI同事协作、任务自动化、聊天辅助

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  AI coworkers that run your work. Self-hostable, AGPL-3.0.

  编报时的 AI 中文解读

  Rerun 是一个可自托管的 AI 同事平台，用户用自然语言描述需求，系统会规划并组建一个 AI 团队，每个成员拥有独立记忆、技能和浏览器，可自动执行任务并在关键操作前征求批准。适用于希望自动化处理邮件、数据分析、社交媒体管理等日常工作的团队或个人。

  编报时的 README 依据

  Install Rerun, the open source AI coworkers app, for me. Follow the guide written for AI assistants: https://raw.githubusercontent.com/rerun-build/rerun/main/INSTALL.md ``` By hand, you need Docker, Node.js 22, and a Supabase project (a free one on [supabase.com](https://supabase.com) works). ```bash git clone https://github.com/rerun-build/rerun.git && cd rerun cp .env.example .env # fill the Supabase URL and keys, and an OpenRouter key node scripts/self-host-check.mjs npx supabase link --project-ref <your-project-ref> && npx supabase db push --include-seed

  docker compose up -d --build # then open http://localhost:3000 ``` That's it, say hi to the Foreman. 🎉 The [self-hosting guide](https://docs.rerun.build/self-host) walks through every step, including Supabase on your own machine, adding servers, HTTPS and updates. ## ☁️ Open source or Rerun Cloud? Same product, same code. Rerun Cloud is this Rerun, hosted, plus what only a hosted service can offer. | | 🛠️ Open source | ☁️ Rerun Cloud | |---|---|---| | Coworkers, chat, memory, skills, schedules | ✅ | ✅ | | Services catalog, MCP, Composio | ✅ | ✅ |

  how to run Rerun from source and how pull requests flow back into Rerun Cloud. Self-hosting is documented, not supported: an issue without logs or steps to reproduce is closed with a link to the guide. Questions and show-and-tell go to [Discord](https://discord.gg/cJUUgdTg5p). If Rerun saves you a few hours, a ⭐ helps a lot. <p align="center"> <a href="https://star-history.com/#rerun-build/rerun&Date"> <picture> <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=rerun-build/rerun&type=Date&theme=dark">

  查看 README 缓存打开 GitHub
• Balanov/shimGo · 430 Stars

  模型服务接入、网页自动化、聊天辅助

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  Tiny local-first AI harness: any model, MCP and Agent Skills, natural-language CLI. Zero dependencies. MIT.

  编报时的 AI 中文解读

  Shim 是一个轻量级的本地优先 AI 工具，通过命令行或自然语言交互，支持连接多种模型、MCP 服务器和 Agent Skills，适合喜欢终端操作的用户。

  编报时的 README 依据

  <a href="https://github.com/Balanov/shim/releases"><img src="https://img.shields.io/badge/platform-Windows%20%C2%B7%20macOS%20%C2%B7%20Linux-lightgrey.svg" alt="Platforms"></a> </p> ## Why Shim? Model-agnostic coding agents are everywhere. Setting them up is still too hard: provider IDs, API endpoints, MCP JSON, skills directories, and a new interface for every tool. **Shim is the small adapter between you and all of that.** - **Talk or type commands.** A full scriptable CLI for experts; ordinary language for everyone else.

  - **Use a local mini-model as the command router.** It translates management requests into an allowlisted command. Mutating commands require confirmation. - **Bring any model API.** Native OpenAI-compatible and Anthropic adapters, plus presets for popular hosted and local providers. - **Bring any MCP server or Agent Skill.** MCP over stdio and Streamable HTTP; lazy `SKILL.md` loading from common directories. - **Stay light.** No Electron, Node, Python, Docker, database, background daemon, or UI framework.

  On PowerShell, use `.\shim.exe` when running the binary from the current directory. ### With Go Go 1.25+: ```sh go install github.com/Balanov/shim/cmd/shim@v0.1.0 shim setup ``` ### From source ```sh git clone https://github.com/Balanov/shim.git cd shim go test ./... go build -trimpath -ldflags="-s -w" -o shim ./cmd/shim ./shim setup ``` Windows also has `scripts/build.ps1`; Unix has `scripts/build.sh`. ## Two ways to use it ### Just talk ```text you › найди локальные модели ↳ /discover --add Run this command? [y/N] y you › show my tools ↳ /tools

  ### Use the complete CLI ```sh # Local models: discover and configure all running servers shim model discover --add # Hosted models shim model add openrouter anthropic/claude-sonnet-4.5 --name claude --use shim model add openai gpt-5 --name gpt --use shim model add huggingface org/model --name hf --use # Any OpenAI-compatible endpoint shim model add custom my-model --name mine \ --base-url http://127.0.0.1:8000/v1 --use # One-shot agent run shim run --read-only "explain this repository" shim run --yes --verbose "fix the tests and verify the result" # Interactive terminal shim ```

  **Shim is not an OS sandbox.** Shell commands and MCP servers run with your user's permissions. Workspace path checks do not contain a malicious process. Use a container, VM, or OS sandbox for untrusted workloads. Treat model output, repository text, skills, and MCP responses as untrusted data. ## Sessions and configuration ```sh shim session list shim session new shim run --resume SESSION_ID "continue" shim status shim doctor shim config path shim config show ```

  you › fix the failing tests in this repository ``` Shim has deterministic offline phrases for common English and Russian management commands. For other languages, assign a local model as the router: ```sh shim router use ollama/qwen3:4b ``` The router sees non-command chat text so it can distinguish configuration requests from agent tasks. Use a loopback model if you want that step to stay on your machine. **No model weights are bundled or silently downloaded**; that would defeat the tiny-binary goal. Your primary model handles ordinary conversation in whatever language it supports.

  查看 README 缓存打开 GitHub
• schematrix/HAS_Prepper_DiskHTML · 425 Stars

  离线知识服务、地图查看、紧急救援参考

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  Raspberry Pi 5 üzerinde çevrimdışı bilgi arşivi: Wikipedia, haritalar ve hayatta kalma kaynakları kendi Wi-Fi ağı üzerinden sunulur, internet gerekmez. Internet-in-a-Box tabanlıdır.Offline knowledge archive on Raspberry Pi 5: Wikipedia, maps and survival resources served over its own Wi-Fi hotspot, no internet required. Built on Internet-in-a-Box.

  编报时的 AI 中文解读

  HAS_PrepperDisk 是一个基于 Raspberry Pi 5 的离线知识存档项目，通过自建 Wi-Fi 热点提供 Wikipedia、地图、医疗和维修指南等内容，无需互联网。它基于 Internet-in-a-Box 和 Kiwix 构建，适合在偏远地区或灾难恢复场景下使用。

  编报时的 README 依据

  <p align="center"><img src="assets/demo.gif" width="400" alt="HAS_PrepperDisk exploded view"></p> **[English](#english) · [Türkçe](#türkçe)** --- ## English An offline knowledge archive on a Raspberry Pi 5. Wikipedia, maps, medical references and repair guides, served over the Pi's own Wi-Fi hotspot. Any phone or laptop that joins the network can open them in a browser. No internet needed. Built on [Internet-in-a-Box](https://github.com/iiab/iiab) and [Kiwix](https://kiwix.org). This repo adds a ready-made configuration, helper scripts and a custom homepage. ### Hardware

  - **Book collections (Water Treatment, Knots, Food for Preppers, Post-Disaster, Medicine) keep loading forever.** These collections store their book list in the browser's local database. iPhone Safari private tabs and the captive portal pop-up window block that. Open them in a normal Safari tab. - **The hotspot never comes up and the log says "invalid WPA passphrase length".** Wi-Fi passwords must be 8–63 characters. Fix `hostapd_password` in `/etc/iiab/local_vars.yml` and `wpa_passphrase` in `/etc/hostapd/hostapd.conf`, then `sudo systemctl restart hostapd`.

  | Part | Notes | |---|---| | Raspberry Pi 5 (4 GB or 8 GB) | | | M.2 NVMe HAT + NVMe SSD (256 GB+) | Content lives here | | microSD card (32 GB+) | Operating system | | 5V/5A USB-C power supply | The official one is recommended | | Active cooler | The Pi 5 and SSD get warm | | Ethernet cable | Recommended during install | <img width="572" height="774" alt="Screenshot 2026-10-30 at 23 22 21" src="https://github.com/user-attachments/assets/0f8cf21f-2407-4346-a9ec-e466cc926ade" /> ### Installation

  **1. Operating system.** Use Raspberry Pi Imager to flash **Raspberry Pi OS Lite (64-bit)** to the SD card. In Imager's settings, enable SSH and set a username and password. Boot the Pi with the SSD attached and connect it to your router with Ethernet. **2. Update and get this repo.** ```bash sudo apt update && sudo apt full-upgrade -y sudo apt install -y git git clone https://github.com/schematrix/HAS_Prepper_Disk.git cd HAS_Prepper_Disk chmod +x install.sh scripts/*.sh ```

  **5. Download content.** Edit `config/content.conf` to choose what to download, then run the script below. The defaults need about 150 GB, and downloads can take many hours. If a download is interrupted, re-run the script and it resumes. ```bash sudo ./scripts/download-content.sh ``` **6. Maps.** Open `http://<pi-ip>/admin` (user `iiab-admin`), go to the maps section of the **Install Content** menu and choose the regions you want. **7. Use it.** Join the Wi-Fi network you created and open **http://box.lan/home**. ### Optional: internet over a USB Wi-Fi adapter

  **3. Mount the SSD at `/library`.** Find the disk name with `lsblk`; it is usually `nvme0n1`. If the disk is not ext4 it will be formatted, and the script asks before doing so. ```bash sudo ./scripts/setup-ssd.sh /dev/nvme0n1 ``` **4. Install.** The script asks for the Wi-Fi name and password, then installs Internet-in-a-Box (30–60 min). **If the Pi reboots or the install stops, run the same command again.** It continues where it left off. ```bash sudo ./install.sh ```

  - **The hotspot doesn't appear.** Connect over Ethernet, check `sudo iiab-diagnostics`, and make sure the install finished: `/etc/iiab/install-flags/iiab-complete` must exist. - **Map downloads fail with "connection reset by peer".** The map server (`contabostorage.com`) is blocked in some countries, e.g. Turkey. Install Cloudflare WARP on the Pi, connect, retry the download in the admin console, then disconnect: ```bash curl -fsSL https://pkg.cloudflareclient.com/pubkey.gpg | sudo gpg --yes --dearmor -o /usr/share/keyrings/cloudflare-warp-archive-keyring.gpg

  查看 README 缓存打开 GitHub
• smarkyie/NSP-BoxPython · 407 Stars

  游戏打包

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  NSP Box: A Switch 2 Game Packager. Experimental Windows tool for lawfully obtained game snapshots and extracted DLC, with reversible emulator patching.

  编报时的 AI 中文解读

  NSP Box 是一款 Windows 便携应用，用于将已提取的 Switch 2 游戏快照打包成 NSP 文件，供兼容模拟器使用。适合拥有合法游戏数据的用户，通过图形界面选择文件夹、添加封面并构建 NSP。

  编报时的 README 依据

  # NSP Box: A Switch 2 Game Packager [Download](https://github.com/smarkyie/NSP-Box/releases/latest) · [Build from source](BUILD.md) · [Changelog](CHANGELOG.md) · [Disclaimer](DISCLAIMER.md) A portable Windows app that packages a complete, already-extracted ExeFS/RomFS snapshot into a plaintext NSP for compatible emulators. Choose a folder, add your own artwork, and build an NSP.

  **Experimental build with limited real-world testing.** If anything goes wrong, [open an issue](https://github.com/smarkyie/NSP-Box/issues/new) with your app/emulator versions, reproduction steps and error messages. Do not upload game files, keys or personal paths. ## Quick start 1. Extract the entire ZIP to a writable folder. 2. Open `NSP-Box.exe`. Python is bundled; installation is not required. 3. Click **Browse folder** and select your main game dump folder, or drag the folder into the window. Pasting a folder path and pressing Enter also works.

  The GUI's export button saves another copy without overwriting anything. Setup runs only when explicitly selected, never during scanning/building. These placeholders support this app's plaintext repacks; they do not replace your own retail keys for encrypted games. Eglantine forks can differ; consult the fork's documentation if its key-loading rules differ. ## Portability and privacy - Windows 10/11 x64; no administrator rights, Python installation, emulator installation, or network connection required.

  The reversible **Emulator patch** panel now lists its supported Windows x64 targets: Selevet performance build 4 (normal/30 FPS GUI), Eglantine nx2 98/96, AMD Experimental 77-2 (build-99), and PokeDinner 2.1.0a/2.0.0. Standard/PGO and GUI/CLI variants are identified explicitly. Compatible Windows x64 Selevet/Eglantine/PokeDinner builds read the metadata inside an NSP, so a separate `main.npdm` beside the game is unnecessary after applying the patch and restarting. See [patch installation, removal and limits](EMULATOR_PATCH.md). Windows x64 downloads only; no Linux build is included.

  The Game tab does not merge separate updates or DLC; DLC folders are left out of game snapshots. Use the DLC tab to create a separate add-on package for installation through a compatible emulator. Multi-program applications, patch-only/BKTR data, encrypted NSP/XCI inputs, compressed ROM packages, decryption, and authenticated cartridge reconstruction are outside this release's scope. This is a snapshot packager, not an emulator or a universal game converter. ## Artwork and metadata

  查看 README 缓存打开 GitHub
• wquguru/shiseiPython · 399 Stars

  姿势参考

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  💃 Free CC0 mannequin pose reference library for AI image tools (Nano Banana, ChatGPT, ComfyUI, Midjourney) — 250+ poses × 5 angles in line art, names in ja/zh/en/ko

  编报时的 AI 中文解读

  Shisei 是一个免费的 CC0 人体模型姿势参考库，包含 280 个姿势、每个姿势 5 个视角的线条画，以及粘土、深度和法线贴图，供 AI 图像工具（如 Nano Banana、ChatGPT、ComfyUI、Midjourney）使用。

  编报时的 README 依据

  **[Open the gallery](https://shisei.indaloop.app/en/) · [Download line art (36.5 MB)](https://github.com/wquguru/shisei/releases/download/v0.4.1/shisei-all-v0.4.1.zip) · [Other downloads](#downloads)** [](data/catalog.json) [](docs/GENERATION.md) [](docs/GENERATION.md)

  description for std, the file for each view, and why a view is empty. ## Using the images Upload a view as the pose or composition reference in your image tool. Using several angles of the same pose keeps the figure consistent across shots. Face, clothing and background come from your prompt or other references. | Tier | Where | Size | Typical file | |---|---|---|---| | `w300` | `images/*/w300/` | 300 px wide | ~7 KB | | `w600` | `images/*/w600/` | 600 × 800 | ~17 KB | | `w900` | `images/*/w900/` | 900 × 1200 | ~29 KB |

  | full | [original zip](#downloads) | 1344 × 1792 (photos 1120 × 2240, redrawn sets 900 × 1200) | ~100 KB | The repository holds the three web tiers: about 145 MB for the poses (line art, photos and the clay, depth and normal maps), plus about 150 MB for the costume library. The full-size originals and ready-made zips are on the [Releases](https://github.com/wquguru/shisei/releases) page; most people only need a zip. ### Downloads

  > **Content note:** the sexy and duo series are suggestive adult pose references (faceless mannequins, no > nudity detail). They stay out of previews and search pages, and the gallery shows them blurred until you tap one > or flip the switch. [](https://shisei.indaloop.app/en/) Shisei is a self-made pose reference library. Every pose is a faceless mannequin drawn from five angles — front,

  from `images/` and the catalog, and `data/qc.json` keeps the reason. - 43 sets were redrawn by a second generator where the original had at most two usable views or showed another pose (catalog `source: digi`, a slightly different mannequin with milder high and low angles; whole sets, never mixed). P038 and P052 still show a different pose from their name and stay hidden. - Views that were never generated stay empty rather than being filled with a bad drawing.

  查看 README 缓存打开 GitHub
• changhaoting1998-star/manufacturing-wechat-miniapp-starterJavaScript · 382 Stars

  生产数据采集、考勤管理、工资确认

  Stars 统计更新于 2026/11/4 07:51:17

  查看原始资料

  作者描述

  Privacy-first WeChat Mini Program starter for manufacturing employee self-service and auditable MES integration.

  编报时的 AI 中文解读

  面向中小制造业的微信小程序启动模板，用于员工现场报工、考勤、工资确认等数据采集，并安全回传MES系统。

  编报时的 README 依据

  Many frontline events are still reported on paper, in chat messages or after a shift. The delay makes production, quality and workforce records incomplete when managers need them. A Mini Program can collect structured data at the point of work without asking employees to install or learn another application, then return validated events to MES, payroll or factory administration systems.

  The intended production model always works with an MES. MES remains the system of record for work orders, processes, routing, equipment, production results and quality records. The Mini Program is the employee-facing collection and feedback layer: it receives authorized tasks from MES and sends verified shop-floor events back through an auditable adapter. This starter focuses on four practical outcomes: - capture production completion, material use, quality, abnormal events, work time and forms as close to the event as possible;

  # Manufacturing WeChat Mini Program Starter [](https://github.com/changhaoting1998-star/manufacturing-wechat-miniapp-starter/actions/workflows/ci.yml) [](LICENSE) A privacy-first, synthetic starter that turns a familiar WeChat Mini Program into a low-friction shop-floor data channel for small and medium-sized manufacturers. ## Why this matters

  cloudfunctions/ Platform-neutral gateway and production boundary src/ Domain rules, memory store and MES connector tests/ Synthetic authorization and failure-path tests examples/ Deterministic local demo ``` ## Mini Program setup 1. Copy `miniprogram/project.config.example.json` to a local `project.config.json`. 2. Replace the placeholder AppID locally. Do not commit the resulting file. 3. Copy `miniprogram/config.example.js` to your own environment-specific configuration.

  4. Connect `cloudfunctions/employeeGateway/handler.js` to a persistent repository and CloudBase identity provider. 5. Configure collections so Mini Program clients cannot read or write employee, payroll, signature, attendance or audit data directly. 6. Compile and preview in WeChat Developer Tools, then verify privacy disclosures before submission. The included cloud gateway is deliberately not deployable without operator-owned identity and storage wiring. This prevents a demo identity model from being mistaken for production authentication. ## Architecture ```text Native Mini Program

  All names, departments, dates, amounts, identifiers and locations in this repository are synthetic. The repository contains no production AppID, CloudBase environment, endpoint, credential, employee record, attendance location, payroll record or signature. ## Quick start Node.js 20 or newer is required for the zero-dependency test and demo runtime. ```bash npm test npm run demo ``` Expected demo output contains only synthetic identifiers, one MES receipt and a SHA-256 payload hash. ## Repository layout ```text miniprogram/ Native WXML/WXSS/JavaScript page skeletons

  │ action + payload ▼ Cloud gateway ── verified platform identity │ ├─ employee access and authorization rules ├─ attendance / payroll / form services └─ MES connector ── idempotency key + payload hash + receipt ``` The client never submits the trusted platform subject identifier. A production adapter must derive it from the verified CloudBase execution context. ## MES integration model This repository provides the boundary pattern; an operator supplies the adapter for its own MES. | Direction | Typical records | Owner | | --- | --- | --- |

  - deployable keys, tokens and environment files; - real employee, organization, payroll, attendance, signature and location data; - direct client access to privileged collections; - automatic background location collection; - a production-ready account, RBAC, database or retention implementation. Before production use, define data minimization, retention, deletion, export, incident response and user notice requirements. Add server-side authorization on every route, encrypted storage, audit retention, rate limits, backups and monitoring. ## Optional Feishu boundary

  项目还可以维护经过审核的员工档案，向本人发送工资单并记录确认结果；基于 MES 已确认的生产事件，在产品、工艺、设备、班次和质量条件可比时分析不同工艺效率，辅助培训、排班和岗位匹配。效率数据只应作为管理者的决策支持，不能单独决定调岗、薪资或去留。正式系统应提供人工复核、记录纠正与申诉机制，并排除与工作无关或受保护的个人属性。 示例数据全部为虚构数据。项目重点演示员工准入、本人考勤、工资确认、动态表单和 MES 回执的安全边界，不包含任何真实企业的小程序标识、云环境、员工、工资、考勤、签名或部署资料。 项目不是可直接上线的完整 MES。正式使用前必须接入企业自己的身份、权限、数据库、审计、加密、备份、数据保留和隐私告知制度。 ## License MIT. See [LICENSE](LICENSE).

  查看 README 缓存打开 GitHub
• mhfed/vn-law-checkShell · 349 Stars

  法律合规检查、代码扫描

  Stars 统计更新于 2026/11/4 07:46:38

  查看原始资料

  作者描述

  Audit pháp lý website theo luật Việt Nam (BVDLCN 91/2025, TMĐT 122/2025, ANM 116/2025, AI 134/2025) cho Claude Code, Cursor, Copilot, Windsurf, Cline

  编报时的 AI 中文解读

  为AI编程助手提供越南法律合规检查的规则集，可扫描网站或移动应用代码，指出违反越南数据保护、电商、网络安全等法律的具体位置，并支持后续修复。适用于使用Claude Code、Cursor等工具的开发者。

  编报时的 README 依据

  /plugin marketplace add mhfed/vn-law-check /plugin install vn-law-check ``` Hoặc clone thẳng vào thư mục skills, không cần qua marketplace: ```bash git clone --depth=1 https://github.com/mhfed/vn-law-check ~/.claude/skills/vn-law-check ``` (muốn cài riêng cho 1 project thì đổi `~/.claude/skills/` thành `.claude/skills/` trong project đó) Dùng Cursor/Windsurf/Copilot/Cline/Codex: chạy lệnh này trong thư mục project: ```bash curl -fsSL https://raw.githubusercontent.com/mhfed/vn-law-check/main/install.sh | bash ```

  Script cài checklist vào `.vn-law-check/` và không ghi đè file nào của bạn. Nếu project đã có `AGENTS.md`, `CLAUDE.md`... thì script chỉ chèn thêm một đoạn ngắn trỏ tới `.vn-law-check/`. Chạy lại để cập nhật cũng không bị chèn trùng. Xong thì bảo agent: "audit pháp lý website này theo luật Việt Nam". Không có source code, chỉ có địa chỉ website cũng được: "audit pháp lý https://ten-mien.vn". Agent sẽ tự quét code để phân loại site trước. Blog hay trang tĩnh không thu thập gì thì chỉ nhận về 1 dòng kết luận, không phải bảng checklist dài.

  Không phải tư vấn pháp lý, xem [DISCLAIMER.md](./DISCLAIMER.md). Luật ở mảng này vừa đổi gần hết trong 2025-2026, nên tự xác minh lại trước khi dựa vào. ### Vì sao làm cái này Rules compliance có sẵn trên GitHub cho AI agent gần như toàn GDPR/CCPA. Chưa có cái nào theo luật Việt Nam, trong khi luật VN vừa thay gần hết: Luật Bảo vệ dữ liệu cá nhân 2025, Luật Thương mại điện tử 2025, Luật An ninh mạng 2025 đều mới có hiệu lực trong 2026, thay các nghị định cũ. ### Dùng với tool nào Cùng một nội dung, đặt ở nhiều file để mỗi tool tự đọc đúng chỗ của nó, không cần config gì thêm:

  Mình là dev, không phải luật sư. Đọc hiểu tới đâu thì viết vào checklist tới đó, chỗ nào chưa chắc thì ghi thẳng "CHƯA XÁC MINH". Đếm lại thì còn 32 chỗ như vậy 😅 Phần lớn nằm ở các nghị định xử phạt (`references/08-penalties-reference.md`) và NĐ 356/2025 (`references/01-personal-data-protection.md`). Nên nếu bạn làm luật, hoặc đơn giản là đọc luật kỹ hơn mình, vào bắt lỗi giúp. Không cần biết code, [mở một issue](https://github.com/mhfed/vn-law-check/issues/new/choose) theo mẫu có sẵn là được: - Thấy chỗ sai: số điều, văn bản hết hiệu lực, mức phạt, cách hiểu chưa chuẩn

  查看 README 缓存打开 GitHub
• drakolordx7/pokopia-recompC++ · 280 Stars

  游戏运行

  Stars 统计更新于 2026/11/4 07:46:41

  查看原始资料

  作者描述

  Pokémon Pokopia (Switch 2) recompiled to run natively on Windows.

  编报时的 AI 中文解读

  将《宝可梦 传说 Z-A》Switch 2 版本重新编译为 Windows 原生运行的工具，需要用户自备游戏文件，适合希望在 PC 上体验该游戏的玩家。

  编报时的 README 依据

  # Pokopia Recomp Pokémon Pokopia (Switch 2) recompiled to run natively on Windows. [Download the latest build](https://github.com/drakolordx7/pokopia-recomp/releases/latest) You need your own extracted copy of the game, version 2.0.1a. Not affiliated with Nintendo, The Pokémon Company or Koei Tecmo. ## Requirements - Windows 10 or 11, 64-bit - GPU with Vulkan 1.3 - 16 GB of RAM ## Usage 1. Unzip the build and run `PokopiaLauncher.exe`.

  2. Pick your dump folder and press Set up. It compiles the game on your PC (about 7 minutes), then builds the shader cache (1 to 2 hours). You can pause it and continue later. 3. Press Play. F1 opens the settings in game. Middle mouse toggles Joy-Con 2 mouse mode for placing items. ## Reporting problems [Open an issue](../../issues/new/choose) with `pkp.log` from the session where it happened, your GPU and your Windows version. ## Contributing - Open pull requests against `main` and say what you tested on which GPU.

  查看 README 缓存打开 GitHub
• Balanov/skeinGo · 260 Stars

  任务编排、紧急停止控制

  Stars 统计更新于 2026/11/4 07:46:41

  查看原始资料

  作者描述

  Fail-closed swarm orchestration with epoch-fenced work and late-completion rejection.

  编报时的 AI 中文解读

  Skein 是一个用于本地代理集群的故障关闭编排工具，执行显式任务图，支持有界并发和持久纪元栅栏，紧急停止会使排队工作失效并拒绝旧纪元的迟到完成。适用于需要严格停止语义的本地多代理工作流。

  编报时的 README 依据

  Both scripts pick the binary for the current OS and architecture (Linux, macOS, Windows × amd64, arm64), verify it against `SHA256SUMS` and install into a user directory. Set `SKEIN_INSTALL_DIR` to override. Alternatively, download binaries from [releases](https://github.com/Balanov/skein/releases/latest) or build from source with Go 1.26+: ```sh go build -trimpath -o bin/skein ./cmd/skein ``` The binary is static, needs no daemon, database or runtime dependencies. ## Run ```sh skein run --state .skein --workers 2 --include-tasks examples/fanout.json

  skein stop --state .skein --reason "manual emergency stop" skein status --state .skein # {"stopped":true,"epoch":2} skein reset --state .skein # inspect first; epoch is never reused ``` A run stores `report.json`, `events.jsonl`, leases and payload data under `.skein/runs/<run-id>/`. Exit codes: `0` success, `3` a task failed, `4` the run was stopped, `1` operational error, `2` usage error.

  [Architecture](docs/ARCHITECTURE.md) · [Security boundary](SECURITY.md) · [Integration](docs/INTEGRATION.md) · [Benchmarks](BENCHMARKS.md) · [Landing source](site/) ## Install Linux and macOS: ```sh curl -fsSL https://raw.githubusercontent.com/Balanov/skein/main/install.sh | sh ``` Windows PowerShell: ```powershell irm https://raw.githubusercontent.com/Balanov/skein/main/install.ps1 | iex ```

  **Not covered:** sandboxing arbitrary commands; authentication or signed leases; semantic validation of agent output; exactly-once external side effects; automatic crash recovery; distributed consensus; remote or detached work, and side effects that already crossed an external boundary. A lease is a fencing token, not a credential — external systems honor it only through a gateway that checks it. Manifest commands keep the OS user's permissions; use an OS sandbox for untrusted workloads. See [SECURITY.md](SECURITY.md). ## Measurements

  查看 README 缓存打开 GitHub
• Balanov/verdictPython · 224 Stars

  工具调用结果验证、模型本地部署

  查看原始资料

  作者描述

  A 197 KB local semantic verifier for AI agent tool calls. Exit code 0 is not proof.

  编报时的 AI 中文解读

  Verdict 是一个极小的本地语义验证器，用于判断 AI 代理工具调用的结果。它读取命令、退出码、标准输出和标准错误，返回 success、failure、blocked 或 unknown 标签，帮助代理区分任务成功、失败、环境阻塞或证据不足。

  查看 README 缓存打开 GitHub
• MagoStudio/MagoStudioTypeScript · 216 Stars

  视频剪辑、视频调色、模型本地部署

  查看原始资料

  作者描述

  MagoStudio application

  编报时的 AI 中文解读

  Mago Studio 是一款本地优先的桌面 AI 视频编辑器，支持脚本转视频、素材重风格化、帧精确编辑等功能。适合希望在本地 GPU 上运行 AI 视频编辑，或使用云模型进行渲染的用户。

  查看 README 缓存打开 GitHub
• xFractals/pokopia-nativeC++ · 188 Stars

  游戏转换

  查看原始资料

  作者描述

  Build Pokémon Pokopia (Switch 2) into a native Mac app and iPhone app from your own dump. Tools only; no game content.

  编报时的 AI 中文解读

  这是一个将用户自己拥有的《宝可梦 传说 Z-A》Switch 2 游戏转换并构建为原生 macOS 和 iOS 应用的工具，供拥有该游戏的用户进行个人使用。

  查看 README 缓存打开 GitHub
• maluramichael/awesome-reverse-engineer-anythingPython · 166 Stars

  逆向工程资源

  查看原始资料

  作者描述

  Trending open projects that reverse engineered apps, games, consoles, protocols and devices, plus the tools they used. Star trends refreshed daily.

  编报时的 AI 中文解读

  这是一个精选的逆向工程项目列表，收录了逆向工程应用、游戏、控制台、协议和设备的开源项目及其使用的工具，并每日更新星标趋势。适合对逆向工程感兴趣的研究者、开发者和爱好者浏览发现。

  查看 README 缓存打开 GitHub
• jwang771/3d-world-simulatorPython · 137 Stars

  3D世界模拟、物理模拟研究

  查看原始资料

  编报时的 AI 中文解读

  这是一个用于多模态3D世界模拟的研究项目，提出了混合粒子-体素动力学方法，并提供了训练和推理代码。适用于研究3D物理模拟和机器人操作的研究人员。

  查看 README 缓存打开 GitHub
• andrewnakas/gtav-recompC++ · 129 Stars

  游戏运行

  查看原始资料

  作者描述

  GTA V (Xbox 360, TU27) static recompilation with ReXGlue

  编报时的 AI 中文解读

  将Xbox 360版《GTA V》静态重编译为原生代码，可在Windows、Linux、macOS、iPhone和Android上运行，需用户自备游戏光盘和TU27更新。

  查看 README 缓存打开 GitHub
• 1337isnot1337/DOPPLERGEISTPython · 128 Stars

  音频定位

  查看原始资料

  作者描述

  A tool to geolocate audio recordings using aircraft Doppler shifts and historical flight tracks.

  编报时的 AI 中文解读

  DOPPLERGEIST 是一款通过分析音频记录中飞机多普勒频移与历史飞行轨迹来推断录音地点的本地运行工具，提供命令行和浏览器上传界面，适用于研究或取证场景。

  查看 README 缓存打开 GitHub
• Open-RUA/uni-robot-agentPython · 127 Stars

  机器人任务运行、模拟环境集成、运行监控与回放

  查看原始资料

  作者描述

  Democratizing Robot-Use Agents: A Unified Harness with Open Models

  编报时的 AI 中文解读

  uni-robot-agent 是一个开源的机器人使用代理的统一工具包，支持 Codex 和 Claude Code 运行时，提供 CLI、Python SDK 和本地仪表盘，用于在模拟环境中运行机器人任务，并支持直接模式和混合模式（如 Pi0.5）。

  查看 README 缓存打开 GitHub
• shashvat1965/slidesenseSwift · 122 Stars

  音量亮度调节、媒体控制

  查看原始资料

  作者描述

  Turn the edges of your MacBook trackpad into volume and brightness sliders. Native macOS menu bar app with a Liquid Glass HUD, haptics and a media strip.

  编报时的 AI 中文解读

  SlideSense 是一款 macOS 菜单栏应用，可将 MacBook 触控板的左右边缘变成音量与亮度滑条，支持触觉反馈、屏幕显示和媒体控制。适合希望快速调节系统音量和亮度的 MacBook 用户。

  查看 README 缓存打开 GitHub
• DominikScholz/open-bot-avatarTypeScript · 121 Stars

  头像生成、头像渲染

  查看原始资料

  作者描述

  A JSON spec and renderer for interactive agent avatars

  编报时的 AI 中文解读

  这是一个用于创建和渲染交互式代理头像的JSON规范和渲染器，提供无依赖的npm包和在线编辑器，适合开发者集成到应用中。

  查看 README 缓存打开 GitHub
• Renly1994/xiaotang-daozaiRust · 110 Stars

  桌面信息展示、音乐歌词显示、邮件验证码提醒

  查看原始资料

  作者描述

  岛仔 Daozai：Rust 构建的 Windows 桌面信息岛。音乐歌词、邮件验证码、日程提醒、待办与 AI 额度，让重要信息轻轻出现在屏幕顶部。

  编报时的 AI 中文解读

  岛仔是一款Windows桌面信息岛应用，将音乐歌词、邮件验证码、日程提醒、待办和AI额度等信息显示在屏幕顶部，适合需要快速查看重要信息的桌面用户。

  查看 README 缓存打开 GitHub
• Toyhom/awesome-computational-character-modelingTeX · 100 Stars

  论文阅读

  查看原始资料

  作者描述

  Survey companion for Computational Character Modeling: role-playing methods, data foundations, and evaluation, organized by the manuscript.

  编报时的 AI 中文解读

  这是一个与计算角色建模综述配套的阅读清单，从行为科学视角整理角色扮演智能体的相关论文，按角色建模、数据基础、评估三大章节组织，适合研究者和学生查阅。

  查看 README 缓存打开 GitHub
• fuyichen0725/blinkmimicPython · 99 Stars

  运动跟踪、策略训练

  查看原始资料

  作者描述

  Whole-body motion tracking for the Unitree G1 in about 3 minutes on one GPU.

  编报时的 AI 中文解读

  BlinkMimic 是一个用于 Unitree G1 人形机器人的全身运动跟踪训练系统，能在单张 RTX 4090 上约 3 分钟内完成策略训练，适用于机器人运动控制研究。

  查看 README 缓存打开 GitHub
• Roland0511/dsh-marktextJavaScript · 98 Stars

  Markdown编辑

  查看原始资料

  作者描述

  MarkText (muya) WYSIWYG Markdown editor for the DeepSeek Harness right sidebar

  编报时的 AI 中文解读

  为 DeepSeek Harness 右侧边栏提供所见即所得 Markdown 编辑器插件，可打开并编辑 .md 文件，支持标题、表格、数学公式、图表和图片，并可直接保存回磁盘。

  查看 README 缓存打开 GitHub
• tmcleroy/soulcalibur2-recompC++ · 96 Stars

  游戏移植

  查看原始资料

  作者描述

  Soulcalibur II (Xbox) as a native port for macOS, Linux/Steam Deck and Windows, with Link and Heihachi from your own discs

  编报时的 AI 中文解读

  这是一个将Xbox版《灵魂能力II》原生移植到macOS、Linux和Windows的非官方项目，支持从玩家自有的光盘中提取角色，并添加了训练模式、在线对战等功能。

  查看 README 缓存打开 GitHub
• 0xCodez/pulsrSwift · 93 Stars

  健康监测、睡眠分析、训练负荷评估

  查看原始资料

  作者描述

  Pulsr — open-source recovery, strain and sleep coach for Apple Watch. Liquid Glass UI, on-device, no subscription.

  编报时的 AI 中文解读

  Pulsr 是一款面向 Apple Watch 用户的开源恢复、训练负荷和睡眠教练应用，利用手表记录的心率、HRV 等数据生成每日评分，并提供个性化建议。

  查看 README 缓存打开 GitHub
• captainj-sparrow/astral-chain-pc-recompC · 88 Stars

  游戏重编译

  查看原始资料

  作者描述

  Static recompilation of Astral Chain to native Windows, with a Vulkan renderer. Game files not included.

  编报时的 AI 中文解读

  将《异界锁链》静态重编译为原生 Windows 应用，使用 Vulkan 渲染器，不包含游戏文件，需要用户自备合法游戏副本。

  查看 README 缓存打开 GitHub
• JDhruv14/ascii-dioramaJavaScript · 88 Stars

  ASCII场景生成、3D场景定制

  查看原始资料

  作者描述

  Claude Code skill for tilt-shift ASCII miniature videos rendered with Three.js

  编报时的 AI 中文解读

  这是一个Claude Code技能，用于创建移轴效果的ASCII微型3D场景，场景由彩色等宽字符构成，可在浏览器中实时运行。适合希望快速生成或定制ASCII艺术风格3D世界的用户。

  查看 README 缓存打开 GitHub
• azmasm/switch2macObjective-C · 82 Stars

  游戏运行、游戏移植

  查看原始资料

  作者描述

  Native Mac builds of Nintendo Switch 2 games on Apple silicon, from your own dump. A fork of nxnative with Switch 2 support (NVN2 on Metal, mesh shaders, movies) and performance work.

  编报时的 AI 中文解读

  在 Apple 芯片 Mac 上原生运行 Switch 2 游戏的工具，需使用用户自有的游戏 dump，将游戏代码和着色器转换为原生运行，非模拟器。适合拥有 Switch 2 游戏 dump 并希望在 Mac 上体验的玩家。

  查看 README 缓存打开 GitHub
• anomalyco/textjsTypeScript · 81 Stars

  控制台应用开发

  查看原始资料

  作者描述

  A React framework for apps that render in the browser console.

  编报时的 AI 中文解读

  Text.js 是一个用于在浏览器控制台渲染应用的 React 框架，提供组件、路由和 CLI 工具，适合开发者构建交互式控制台界面。

  查看 README 缓存打开 GitHub

其余 8 个项目见收录列表。

查看本期全部收录项目