跨设备协同、AI视频特效与自托管工具：今日开源项目里的实用新选择

收录数按本期首次达标收录的项目统计。

• InukaAsith/TwinDeskJavaScript · 1,296 Stars

  跨设备协同、屏幕共享、文件传输

  Stars 统计更新于 2026/10/30 07:51:28

  查看原始资料

  作者描述

  All your computers and your phone, working as one desk. Screens, apps, files, clipboard, calls, texts, notifications, mouse and keyboard, shared between Mac, Windows, Linux and Android. Free and open source. No account, no cloud, and now no shared network needed.

  编报时的 AI 中文解读

  TwinDesk 是一款免费开源的跨设备协同工具，让 Mac、Windows、Linux 和 Android 设备共享屏幕、应用、文件、剪贴板、鼠标键盘、通知和通话，无需账号或云端。

  编报时的 README 依据

  <img src=".github/assets/hero.jpg" alt="One Mac's extended screen in a window on another Mac, the Mac's desktop extended onto a tablet, and TwinDesk on a phone" width="100%"> </p> Your devices are good at their own jobs and bad at working together. TwinDesk fixes that. Put a **Mac and a Windows PC** side by side and use them with one mouse and keyboard. Open the PC's apps on the Mac. Turn an old laptop or your **Android tablet** into a second screen. Answer your **phone's calls and texts** from the computer, use its **camera in a video call**, and paste

  ### 💻 [Computers](#-computer-to-computer) One mouse and keyboard · second screen · the other computer's apps in windows · files · clipboard · **camera** · sound · terminal </td> <td width="33%" valign="top"> ### 📱 [Phones and tablets](#-phone-and-tablet) Calls · texts · photos · files · **camera** · **remote and gamepad** · screen and apps in windows · tablet as a second screen </td> <td width="33%" valign="top"> ### 📡 [Connections](#-connections) Same Wi-Fi · USB cable · **Bluetooth** · **Wi-Fi Direct** and **hotspot** links made on demand · picked for you, moved for you </td> </tr>

  [How it works](#how-it-works) · [build and run](#build-and-run) on macOS, Windows, Linux, NixOS and Android · [GitHub Actions](#github-actions) </td> </tr> </table> <a name="download"></a> ## ⬇️ Download <table> <tr> <td width="25%" align="center" valign="top"> **🍎 macOS**<br>_{13 Ventura or later} <a href="https://github.com/InukaAsith/TwinDesk/releases/download/v2.4.0/TwinDesk-2.4.0-arm64.dmg"><img src="https://img.shields.io/badge/Apple_Silicon-.dmg-000000?style=for-the-badge" alt="macOS, Apple Silicon"></a><br>

  _{The full app adds notifications, texts and Accessibility; the computer can install it for you over wireless debugging} </td> </tr> </table> <p align="center">🍺 <b>Homebrew</b> on macOS: <code>brew install --cask inukaasith/twindesk/twindesk</code><br> _{Update with <code>brew upgrade --cask twindesk</code>: TwinDesk opens again by itself if it was running, so a Mac you drive only through it gets its mouse back. The <a href="https://github.com/InukaAsith/homebrew-twindesk">tap</a> follows every release.}</p> <a name="first-launch"></a> <details>

  <a href="https://twin-desk.vercel.app"><b>Website</b></a> · <a href="#-new-in-23"><b>What's new</b></a> · <a href="#-get-started"><b>Get started</b></a> · <a href="#how-it-works"><b>How it works</b></a> </p> <p align="center"> <img alt="Version" src="https://img.shields.io/github/v/release/InukaAsith/TwinDesk?label=version&color=4b8bff"> <img alt="macOS 13+" src="https://img.shields.io/badge/macOS-13%2B-black"> <img alt="Windows 10/11" src="https://img.shields.io/badge/Windows-10%20%7C%2011-0078d4">

  <img alt="Linux X11 and Wayland" src="https://img.shields.io/badge/Linux-X11%20%7C%20Wayland-f0a30a"> <img alt="Android 10+" src="https://img.shields.io/badge/Android-10%2B-3ddc84"> <img alt="License: GPL-3.0" src="https://img.shields.io/badge/license-GPL--3.0-blue"> <a href="https://github.com/inukaasith/twindesk/releases"><img src="https://img.shields.io/github/downloads/inukaasith/twindesk/total" alt="Downloads"></a> </p> <p align="center">

  <summary><b>⚠️ First launch: the apps aren't code-signed</b> _{how to allow them on macOS, Windows and Android}</summary> TwinDesk is free and has no paid Apple Developer ID or Windows code-signing certificate, so macOS and Windows don't know who made it and ask once before opening it. The source is all here. **🍎 macOS** The easiest way is **Homebrew**, which installs it ready to open: `brew install --cask inukaasith/twindesk/twindesk` From the .dmg instead: 1. Drag TwinDesk into Applications and open it. macOS says it can't verify TwinDesk: click **Done** (not Move to Bin).

  - **One notification pool.** Your phone buzzes, and every computer shows it. - **Pair once, join all.** A new device paired with one computer meets the rest by itself. Mesh is in **beta** on Windows, Mac and Linux: turn it on in the **Multi-device** tab, and off again any time with everything as you left it. Connecting just two devices? You won't need it. Still coming to Mesh: the Camera, pictures on the clipboard, and files dragged to more than one device at a time; the phone connects to one computer at a time, and Linux joins over the same Wi-Fi only. --- # 📱 Phone and tablet

  查看 README 缓存打开 GitHub
• oscar-soirey/RavenC · 1,116 Stars

  GUI界面构建

  Stars 统计更新于 2026/10/30 07:55:44

  查看原始资料

  作者描述

  A lightweight, multithreaded GUI library in C for editors and tools.

  编报时的 AI 中文解读

  Raven 是一个轻量级、多线程的 C 语言 GUI 库，专为编辑器和工具软件设计。它提供类似 Slate 的布局系统，支持 XML 定义界面，并通过官方 IDE、多语言绑定等功能，帮助开发者高效构建跨平台桌面应用。适合需要高性能、现代界面效果的开发者。

  编报时的 README 依据

  # Raven A lightweight, multithreaded GUI library in C for editors and tools. ## Disclaimer : This project is still in active development and public API is subject to change. ## Build from source Raven builds with CMake and a C11 compiler, and has no dependency to install besides the X11 development package on Linux. The Vulkan *headers* are downloaded automatically during the configuration (git and an internet connection are needed for that), unless the [Vulkan SDK](https://vulkan.lunarg.com) is installed. No Vulkan library is linked.

  The build trees are in `build\dist\`. `./dist.sh` makes the Linux packages. `cmake --build build --target package` makes a simpler archive (one library) with CPack. ## Examples  *`examples/bevy_editor`: the viewport of the scene editor written in `.xmlrs`, with the move gizmo on the selected cube (here drawn by its software renderer).* | | | |---|---| | `examples/hello` | a window with the C API |

  `.xmllua`. `ravenc` compiles the XML into your language, so everything is checked by your compiler and built by your usual tools (CMake, Cargo, `go generate`, MSBuild, `zig build`, a Python or Lua import). See [XMLC](#one-file-any-language-xmlc). - **The official IDE** (`raven_designer`): new projects in all eight languages, interfaces edited visually next to their code, completion of the Raven API, translations in a table, build, run and compiler errors at their line in the `.xml…` file. See [The IDE](#the-ide).

  | `.xmlpy` | Python | `import raven_xmlpy`, then `import main` or `python -m raven_xmlpy main.xmlpy` | `def f(sender, action)` | `ui.x` | | `.xmlgo` | Go | `//go:generate ravenc main.xmlgo -o main.xmlgo.go` | `func f(raven.ID, string)` | `ui.x` | | `.xmlcs` | C# | MSBuild: `bindings/csharp/Raven.XmlCs.targets` | `static void F(uint, string)` | `MainUi.x` | | `.xmlzig` | Zig | `build.zig`: `xmlzig.compile` (`bindings/zig/xmlzig.zig`) | `fn f(Rid, ?[*:0]const u8, ?*anyopaque) callconv(.C) void` | `ui.x` |

  find_package(raven REQUIRED) # configure with -DCMAKE_PREFIX_PATH=<path to install> target_link_libraries(app PRIVATE raven::raven) ``` or, without installing, `add_subdirectory(Raven)` before the same `target_link_libraries`. Prebuilt packages for every platform are attached to the [releases](https://github.com/oscar-soirey/Raven/releases). - **The interface and its code in one file, in your language.** Write the window in XML and its code in `<script>` beside it: `.xmlc` (C), `.xmlc++`, `.xmlrs` (Rust), `.xmlpy`, `.xmlgo`, `.xmlcs` (C#), `.xmlzig`,

  Rslot_set_halign(box, R_HALIGN_CENTER); Rslot_set_valign(box, R_VALIGN_CENTER); Rwindow_set_content(window, box); Rwidget_add_child(box, Rlabel_create("Hello from Raven")); Rid button = Rwidget_add_child(box, Rbutton_create("Quit")); Rbutton_set_clicked_callback(button, on_clicked, NULL); while (Rwait_frame(-1.0)) {} /* the loop is yours: see "The loop" */ Rshutdown(); return 0; } ``` The same window in XML (`Rwindow_create_from_xml_file("hello.xml")`): ```xml <Window Title="Hello" Width="420" Height="220">

  the language's own tools (`CMakeLists.txt`, `Cargo.toml` and `build.rs`, `go.mod`, `.csproj`, `build.zig`) and `strings.csv`. The **Explorer** shows the files; a double click opens each one where it belongs. **Interfaces, visually, next to their code.** A `.xml`, or the interface of a `.xml…`, opens in the visual editor, the same document shown five ways and kept in step: - **Widgets**: what can be added, the tags of the library and those the document declares with `<Widget>`. A double click adds inside the selection.

  Each build below gives the library, `ravenc` (the XMLC compiler), the designer, the examples and the tests. Add `-DRAVEN_BUILD_SHARED=ON` to the configure command for a shared library (`raven.dll`, `libraven.so`, `libraven.dylib`) instead of a static one. ### Windows (Visual Studio) 1. Install [Visual Studio 2022](https://visualstudio.microsoft.com/downloads/) (Community is fine) with the **Desktop development with C++** workload; it includes CMake. Install [Git](https://git-scm.com/download/win). 2. Open the **Developer PowerShell for VS 2022** and run: ```powershell

  ctest --test-dir build --output-on-failure # optional: the tests cmake --install build --prefix install # or: sudo cmake --install build ``` The programs are in `build/bin` (`./build/bin/raven_editor`, `./build/bin/raven_designer`). At run time the Vulkan driver of the system is used (`libvulkan.so.1`, from the `libvulkan1` / `vulkan-loader` package). ### macOS 1. Install the Xcode command line tools (`xcode-select --install`) and CMake (`brew install cmake`, or the app from [cmake.org](https://cmake.org/download/)). 2. Build: ```sh

  real GPUs, Windows, macOS and Linux (X11), the screen readers of each system, `raven_web`, the designer, the build scripts, `dist.sh`/`dist.bat`, the CMake package and every binding (C++, Python, Rust, Go, C#, Zig, LuaJIT). ## Build options and scripts Requirements: CMake 3.16+, a C11 compiler. Linux also needs the X11 development package (`libx11-dev`). The Vulkan *headers* are needed to build; they come from the Vulkan SDK when it is installed and are downloaded otherwise. No Vulkan library is linked. ```sh cmake -S . -B build -DCMAKE_BUILD_TYPE=Release

  or a modified copy of it, keep its license and copyright notices and say what you changed. Contributions (issues, pull requests) are welcome on [GitHub](https://github.com/oscar-soirey/Raven). Questions: oscarsoirey.contact@gmail.com. ## Status Version 0.1. Everything in this README has been tested: the headless test programs (layout, widgets, XML, threads, memory, animations, text editor, shell, dialogs, themes, Markdown, tools, locales and accessibility, clean under AddressSanitizer, UBSan and ThreadSanitizer), the renderer and glass on

  查看 README 缓存打开 GitHub
• c1overzz/dsh-bili-scrollJavaScript · 1,016 Stars

  视频筛选、日报生成、自动提醒

  Stars 统计更新于 2026/10/30 07:51:28

  查看原始资料

  作者描述

  DSH 插件：每天替你刷一轮 B站，读字幕挑出值得的、写一份日报，值得的 @ 你

  编报时的 AI 中文解读

  这是一个DSH插件，用于自动浏览B站视频，通过阅读字幕内容筛选值得关注的视频，并生成日报，可自动评论或私信提醒用户。适合希望高效获取B站内容、减少手动浏览时间的用户。

  编报时的 README 依据

  > ✅ **已经做到了**（v0.1.1 起）：插件会**自己把 `uv` 下到自己的家目录** ⇒ 这条前提不用用户操心、也不用管理员权限。 > ⚠️ 万一它自己也下不动（有用户反馈桌面版上这条链路经常不通）：**手动放一份 `uv.exe` 到 `<家目录>\bin\`** 就行 —— > 插件会**用那一份、跳过下载**（你的 `PATH` 里本来就有 `uv` 的话，它直接用那个）。 ## 装 > 🔴 **先看这条：npm 上还是旧版（0.1.7），别用 `add dsh-bili-scroll` 装。** > 我们的 npm 发布账号还在申诉里（工单 4880295），2026-10-17 之后的新版本**只能**从下面两条装。 ```bash # ⭐ 首选：Release 里的 tgz 直链（不需要 git，也不会被 pnpm 拦构建脚本） dsh plugin --profile web add https://github.com/c1overzz/dsh-bili-scroll/releases/download/v2.0.2/dsh-bili-scroll-2.0.2.tgz # 备选：git 方式（需要机器上装了 git；pnpm 10 可能拦它的构建脚本）

  ⚠️ 我们自己也只在 **npm 0.1.7-rc.x + web profile** 上实测过；桌面版这条路靠用户反馈修好的 —— 装完还是不行的话，把 `duty.log` 里那几行发来（里面会写清楚"宿主依赖是从哪儿取的"）。 ## 首次使用（两步） 1. **装主依赖** —— 设置页点「装主依赖」：在插件自己的家目录建一个独立 venv，装 4 个包 （`bilibili-api-python` / `aiohttp` / `mcp<2` / `qrcode`）。**走国内镜像 + uv，通常几十秒**（进度写 `install.log`，不卡页面）。 2. **扫码登录** —— 点「扫码登录」，用手机 B站 扫页面上那张码。 ⚠️ **确认时选「记住我 / 长期」那张**（选 24 小时的话，明天就得重扫）。 凭证只落在你本机的插件家目录：`<DSH_HOME>/bili-duty/bili_credential.json`，不上传任何地方。 ## 设置页里有什么 分成四页签，第一次用建议只看「日常」： | 页签 | 里面有什么 | |---|---|

  # dsh-bili-scroll ｜ 刷B站 > **先看三行** > - 需要 **dsh ≥ `0.1.7-rc.1`**（本机是 `0.1.7-rc.2`）。⚠️ 版本不满足时，**内核会静默跳过这个插件** —— 装了跟没装一样，不会报错。 > - **空间**：不装「本地转写」约 **84 MB**；装了转写约 **1 GB**（其中 Whisper 模型 538 MB，可在设置里换更小的模型）。 > - 装完**必须重启一次** dsh（插件在 bundle 层组装）。 > - ⚠️ **桌面版 DSH（NEXT / 0.2.x）**：**0.1.3 及以前的版本在桌面版上装不起来** —— 插件整条不激活， > DSH 只留一句 `warning: 1 entry did not activate … failed to import`（根因是宿主半 import 的两个包 > 在桌面版 profile 里没被安装，详见下面「装不上」那节的第 5 条）。**0.1.4 起已修** ✓。 > 另有用户反馈桌面版上「自动跑那一轮会卡住」—— 卡住就先只用设置页的「**现在跑一轮**」手动跑（那条稳）。 > > ⚠️ **装的是旧版吗？** 如果设置页里那个按钮写的是「**重装依赖**」（而不是「**重装主依赖**」），说明你装的是 0.1.0 ——

  - 可以在**峰谷时段自动跑**（DeepSeek 的优惠时段），省 token ## 装之前（一条前提，30 秒） 机器上要有 **`Python 3.10+` 或 `uv`** —— **有一个就够**： - **有 `uv` 最好**：单个 ~32 MB 的 exe，**解压就能用、不用管理员、不用进 PATH**，而且它还能自己把 Python 拉下来； - 只有 Python 也行（装官方安装器时记得勾 **`Add to PATH`**）； - ⭐ **两个都没有？也没关系**：点「装依赖」时**插件会自己下一份 `uv` 放到自己的家目录** （单文件、约 15MB、**免管理员、不用进 PATH** ✓），之后全程自动 ✓ · 它会**先直连 GitHub，不通就走国内加速代理**，还配了 PyPI 镜像兜底 ✓ · 下载完会**核对官方 `sha256`**（对不上就丢弃换下一个源 ✓）—— 因为它下来的东西是要**被执行**的 · ⚠️ **如果杀软 / Windows Defender 报这个下载**：把插件家目录 `%USERPROFILE%\.dsh\bili-duty\bin\` **加白**即可； 不想加白也行，自己装一份 uv 或 Python 3.10+ 就绕过去了 ✓

  dsh plugin --profile web add github:c1overzz/dsh-bili-scroll ``` > 连不上 github 的话，在链接前面加镜像：`https://gh-proxy.com/`（实测可用，返回的是同一个文件）。 > 版本号那一节会变，最新的下载地址在 **Releases** 页（`…/releases/latest`）。 （⭐ 最省事的一条：把仓库链接贴给你自己的 dsh，说「**帮我装这个插件**」，它会自己装。） ⚠️ 装完**必须重启一次 dsh**（bundle 层要重启才组装）。然后进 **设置 → 刷B站**。 ## 装不上 / 卡住时，先看这五条（都是用户实测撞过的） 1. **桌面版（DSH NEXT / 0.2.x）上"自动跑"会卡住** —— 有用户反馈：它起那一轮用的是 `dsh --profile headless …`，桌面版上容易卡在那儿。⇒ **先用设置页的「现在跑一轮」手动跑**（那条是稳的）； 能不能自动跑，等你确认过自己那台桌面版的版本与 profile 再说。 2. **「装依赖」卡住、或者扫码时一直不出二维码** —— 大概率是 0.1.0 那版（见开头那个按钮指纹，跑一次 `add` 就升级了）。

  查看 README 缓存打开 GitHub
• qybaihe/blaTypeScript · 728 Stars

  软件防护

  Stars 统计更新于 2026/10/30 07:40:42

  查看原始资料

  作者描述

  🛡️ BLA (Blind LLM Agents) — Adversarial defense shield that protects binaries, Electron apps, JS bundles, Python, and WebAssembly against AI reverse-engineering agents like REA. Poison -> Exhaust -> Hallucinate.

  编报时的 AI 中文解读

  BLA 是一款防御性工具，用于保护开发者自己的软件（如二进制文件、Electron 应用、JS 包、Python 和 WebAssembly）免受 AI 逆向工程代理（REA）的攻击，通过投毒、耗尽和幻觉机制使逆向分析失效，同时保持运行时行为不变。

  编报时的 README 依据

  [Quick start](#quick-start) · [Empirical results](#empirical-results-vs-real-rea) · [Format coverage](#format-coverage) · [Benchmark](#benchmark-14-targets-internal-estimates) · [FAQ](#faq) ```bash bla harden . # or tell your agent: "Harden my software against REA" ``` </div> > [!IMPORTANT] > BLA is a **defensive** tool for protecting **your own** software. See [DISCLAIMER.md](DISCLAIMER.md). > Get BLA from [qybaihe/bla](https://github.com/qybaihe/bla) or [npm `bla-agents`](https://www.npmjs.com/package/bla-agents). <div align="center">

  | **Hallucinate** | Leave the analyzer with truncated inventories, redaction failures, and ambiguous rabbit holes — while **runtime stays equivalent**. | One sentence covers the common case (auto-discovers `dist/` / `build/` / `.asar` / `.wasm` / `.sqlite` / models / … and runs audit → protect → verify): ```bash npx bla-agents harden . # MCP / agent: bla_harden · "Harden my software against REA" ``` Same pipeline as `bla defend` / `bla_defend` when you already know the path; `harden` is the default one-command UX. ## Quick start ```bash

  # Node.js ^22.23.4 or 24.x recommended (REA red-team needs zlib.zstd on 22.23+) npx bla-agents setup # register MCP + skill with your coding agent npx bla-agents harden . # one-command Counter-REA pass on the current project npx bla-agents benchmark # 14-target showcase (local estimates + runtime checks) ``` Ask your agent after setup: ```text Harden my software against REA ``` Or run stages explicitly: `audit` · `protect` · `verify` · `defend` — see [CLI and MCP](docs/cli-and-mcp.md). --- ## Empirical results vs real REA

  Recorded under **Node v24.22.1** with optional **Wabt 1.0.42** + **tshark 4.4** when noted. Soft-skips keep Node 22 / CI green when tooling is absent. | Format | Before → after | REA outcome after BLA | Budget / error | | --- | ---: | --- | --- | | **Go buildinfo** | 50 KiB → 1.16 MiB | capacity error | `MAX_BUILD_INFO_BYTES` (1 MiB) — *Go build-info string length exceeds the 1 MiB limit* | | **Erlang BEAM** | 84 B → 262 KiB | capacity error | `max_chunks` — *BEAM form exceeds the 64-chunk budget* |

  ### Designed-for but **not** empirically verified here These shields exist in-tree and are covered by unit tests / catalog drift checks, but the matching REA path needs **native tooling we did not run** on this box: | Area | Why unverified here | | --- | --- | | Ghidra / native decompilation depth | Requires a Ghidra install + REA bridge | | Frida / LLDB / GDB / Rizin live sessions | Requires those debuggers + device/process attach | | JADX / full Android APK pipeline | APK resigning unsupported; DEX shield is in-tree |

  | **Firmware** containers | Skipped — staging/extraction budgets exist; no honest E2E without firmware corpora | | **SPIR-V / CUDA fatbinary / AMD GPU objects** | Skipped — need native GPU object decoders; **PTX text** is shielded in-tree (benchmark only, not a REA GPU-object budget hit) | | EVMole in-process EVM decode | Documented as not pathable in-process; EVM shield still densifies to the 4 MiB input approach | We do **not** claim those paths are empirically bankrupt until `redteam:rea` (or an equivalent harness) records a real before/after against the live analyzer. ---

  查看 README 缓存打开 GitHub
• zeabur-oss/zeaburRust · 503 Stars

  服务部署、集群管理、服务诊断

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  Self-hosted PaaS on your own Kubernetes.

  编报时的 AI 中文解读

  Zeabur 是一个自托管的 PaaS 平台，用于在您自己的 Kubernetes 集群上部署和运行服务。它提供 Web 界面来管理项目、服务、数据库和集群，并以通俗语言提供诊断信息。 该平台面向希望在自己的 Kubernetes 集群上获得类似 PaaS 体验的用户，提供项目视图、服务诊断、数据库连接信息和集群状态监控等功能。

  编报时的 README 依据

  <picture> <source media="(prefers-color-scheme: dark)" srcset="docs/images/cover-dark.avif"> <img alt="Zeabur: deploy like a PaaS, get standard Kubernetes; the review of a new service lists a Deployment, a Service and an Ingress with their YAML before it creates them" src="docs/images/cover-light.avif"> </picture> # Zeabur **Deploy and run your services on your own Kubernetes, simply and securely.** Zeabur is an open-source platform you host yourself. It runs in your cluster (or on your

  machine), shows the cluster as projects and services, explains what is wrong in plain words, and lets you change things safely. It works with the objects you already have, never acts with more rights than you have, and keeps no state of its own: no database, no custom resources. **[oss.zeabur.com/zeabur](https://oss.zeabur.com/zeabur/)** shows what Zeabur does, and **[the documentation](https://oss.zeabur.com/docs/zeabur/)** covers installing, deploying, configuring and operating the cluster, and leaving Zeabur, in English,

  - **One Linux server**: the installer sets up K3s (or RKE2), Zeabur Identity and Zeabur. [Install on a single server](https://oss.zeabur.com/docs/zeabur/get-started/quickstart/) - **An existing Kubernetes cluster**: install the Helm chart, with sign-in through Zeabur Identity. [Install in an existing cluster](https://oss.zeabur.com/docs/zeabur/get-started/install/) - **Just you, with your kubeconfig**: `zeaburctl ui` opens Zeabur in your browser on your own computer. [Personal use](https://oss.zeabur.com/docs/zeabur/get-started/personal/) and

  [zeaburctl](https://oss.zeabur.com/docs/zeabur/reference/zeaburctl/) [Releases](https://oss.zeabur.com/docs/zeabur/reference/releases/) lists what a release contains and how to verify it. ## Build from source In the [Nix](https://nixos.org) dev shell, which provides every tool: ```sh git clone https://github.com/zeabur-oss/zeabur && cd zeabur nix develop ./scripts/build-ui.sh ZEABUR_UI_DIST="$PWD/web/dist" cargo build --release --locked -p zeaburctl ./target/release/zeaburctl ui # --context <name> for another context ```

  [繁體中文](https://oss.zeabur.com/zh-TW/docs/zeabur/) and [简体中文](https://oss.zeabur.com/zh-CN/docs/zeabur/). > **Status: pre-alpha (0.1.1).** The API may still change, and it has been tried on few > clusters. Do not point it at a cluster you cannot afford to break. <table> <tr>

  查看 README 缓存打开 GitHub
• Jestions-dev/BlueFatFishJavaScript · 363 Stars

  新标签页美化、自定义壁纸、自定义标签

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  蓝色大肥鱼的新标签页：蓝底玩具箱 + 自定义壁纸/标签，纯本地，断网也能用

  编报时的 AI 中文解读

  这是一个浏览器新标签页扩展，提供蓝底玩具箱背景和可自定义的壁纸、标签，纯本地运行，断网也能使用。适合喜欢个性化浏览器起始页的用户。

  编报时的 README 依据

  - **玩具箱背景**：蓝底 + 随机 5~12 条大肥鱼。无重力、完全弹性、有空气阻尼所以会慢慢停下；撞墙按入射角反弹；可以拖起来甩，拖动时会把别的鱼撞飞（传冲量）；每次刷新随机挑**一条**能被捏——短按松手会「叽」一声（小黄鸭音效）并冒泡，按住不放就是拖。 - **自定义壁纸**：本地图片 / GIF / 动图 / 视频，或者直链、任意网页、B 站 / YouTube 视频页（自动转嵌入播放器）。 - **自定义标签**：自动抓站点图标（页面 `<link rel=icon>` → `/favicon.ico` → 主域名兜底，缓存到本地）；抓不到就用文字图标，最多 8 个字、字号自适应；也可以上传自己的图当图标。 - 时钟 / 日期可关；压暗、模糊滑杆。 - 数据全在本机，没有任何网络请求（除了首次抓图标）。 ## 安装 从 [Releases](https://github.com/Jestions-dev/BlueFatFish/releases) 下载 `BlueFatFish-x.y.z.zip` 或 `.crx`。 **方式一：解压加载（最稳）** 1. 把 zip 解压到一个不会随手删掉的位置 2. 打开 `edge://extensions/`（Chrome 是 `chrome://extensions/`）

  3. 打开 **开发人员模式** 4. 点 **加载解压缩的扩展**，选中解压出来的文件夹 5. `Ctrl+T` 新开标签页 **方式二：装 crx** 1. 同样先打开 **开发人员模式** 2. 把 `.crx` 拖进扩展页面，确认安装 > Edge / Chrome 对非商店来源的 crx 有拦截，装不上就用方式一，效果一样。 用 crx 装和用文件夹装出来的扩展 ID 不同，配置互不相通（一边的标签不会自动出现在另一边）。 ## 使用 右下角齿轮打开设置。 - **壁纸**：选文件 / 贴链接 / 点「玩具箱背景」 - **标签**：设置里增删改，用 ↑↓ 调顺序；图标三种模式（自动 / 文字 / 上传） - **显示**：时钟、日期；「自定义壁纸上也放鱼」（默认关，打开后数量自动少 2 条）、鱼数量滑杆（0 = 自动，按屏宽算） ## 数据存哪 | 内容 | 位置 | | --- | --- | | 配置（壁纸、标签、开关） | `chrome.storage.local` 的 `config` | | 壁纸文件、上传的图标、抓到的站点图标 | IndexedDB `bluefatfish` | ## 给未来的 DSH 插件留的接口

  回复统一是 `{ ok, protocol: 'bff/1', ... }`，出错是 `{ ok: false, error }`。 ## 目录 ``` manifest.json MV3 清单 background.js 对外消息桥 newtab.html 页面 assets/app.js 主逻辑 + 设置面板 assets/store.js 存储 + 抓图标 assets/wallpaper.js 壁纸引擎 assets/toys.js 玩具箱物理 assets/style.css 样式 assets/toys/ 表情包（157 个 WebP + index.json） assets/sfx/ 小黄鸭音效 icons/ 扩展图标 ``` ## 已知限制 - **网页当壁纸**：不少站点发 `X-Frame-Options`，嵌不进去只能白屏；B 站 / YouTube 走官方播放器 iframe 没问题。 - **自动抓图标**首次需要联网，抓完就缓存。 - 少数表情包原本带底色（彩底、黑底），抠不掉。 ## 素材

  查看 README 缓存打开 GitHub
• Vibha-Ramprakash/claude-video-effectsPython · 356 Stars

  视频特效、视频剪辑

  Stars 统计更新于 2026/10/30 07:46:39

  查看原始资料

  作者描述

  36 video effects for Claude Code: drop in a talking-head clip, type /effects orbit. Face tracking, behind-your-head layers. By @vibha.in.progress

  编报时的 AI 中文解读

  为Claude Code提供45种视频特效的Skills合集，用户将视频拖入对话并输入特效名称即可应用，适合内容创作者快速制作带特效的短视频。

  编报时的 README 依据

  > Not sure how? Open Claude Code and paste: **"Install the skill from github.com/Vibha-Ramprakash/claude-video-effects into ~/.claude/skills/effects"**. Claude will do it for you. ## Use it 1. Open Claude Code (the app or the terminal). 2. Drag your video into the chat. 3. Type the effect you want: ``` /effects orbit from 3 to 6 seconds /effects spotlight when I say "kill the lights", with the word Claudit behind me /effects comment-box at the end, typing CLAUDE /effects glass-widgets for the first 3 seconds ```

  The very first time, Claude sets itself up (about 2 minutes). It needs `ffmpeg` and Python, and it installs them for you if they're missing. Claude checks a few frames first, fixes anything that covers your face, then renders `yourvideo_effects.mp4` right next to your original. Your audio stays as it was. You can stack effects in one go: *"Add locked-on from 6 to 10s, then orbit from 10 to 13s, then a reject stamp on 'no'."* ## The 45 effects | # | Type this | What happens | |---|---|---| | 1 | `noise` | Reels pile up behind your head |

  - Charts, counters and headlines should use your real numbers. Tell Claude what they are. - It renders in 1080p. Ask for 4K if you want it (slower). ## Honest notes - These are **starter versions**. They won't match the exact edit in my reel, which also has sound design and hand-tuned timing. - It drafts, you approve. Watch the result before you post. - No sounds or captions are added. Add those in CapCut. - My full edit style is coming later. Follow [@vibha.in.progress](https://instagram.com/vibha.in.progress) to hear first. ## If something goes wrong

  查看 README 缓存打开 GitHub
• Krusty84/yaDisk-HealthPython · 345 Stars

  文件可用性检查

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  Проверка доступности ваших файлов на Yandex Диск

  编报时的 AI 中文解读

  yaDisk-Health 是一个用于检查 Yandex Disk 文件可下载性的工具，通过读取文件开头字节来验证，并提供图形界面和命令行两种使用方式。

  编报时的 README 依据

  - Попытка скачивания интересующего файла через веб-интерфейс Yandex Диска. ## Для кого этот проект? Для пользователей Яндекс Диска, которые хотят найти файлы с ошибками скачивания, повторно проверить их доступность или сохранить отчёт. ## Как установить и использовать ### 1. Что понадобится - Учётная запись Яндекса с файлами на Диске. - Подключение к интернету. - Python 3.10 или новее — программа для запуска скрипта. Ниже для Windows и macOS используется Python 3.13.

  - Загрузка списка файлов, включая вложенные папки, без содержимого корзины. - Вы можете использовать графический интерфейс или терминал (командная строка) в Windows, Ubuntu и macOS. - Проверка всего сохранённого списка, ранее недоступных файлов или случайной части ранее доступных. - Проверка выбранных файлов. - Показ результата, времени проверки, объёма полученных данных и причины ошибки. - Сохранение истории отдельных файлов и итогов каждого запуска. - Экспорт текущего состояния (перечень файлов на Диске, их состояние) в CSV формат.

  - Библиотека `requests` — компонент для сетевых запросов. Команды установки приведены ниже; зависимость `urllib3` установится вместе с ней. - Tkinter — компонент Python для оконного интерфейса. База данных SQLite входит в Python. Отдельный сервер базы данных не нужен. Приложение Яндекс Диска для компьютера также не требуется. Команды вводите в терминале (командной строке) по одной строке и нажимайте Enter. Если команда завершилась с ошибкой, сначала устраните её. ### 2. Скачайте проект 1. Откройте [yaDisk-Health](https://github.com/Krusty84/yaDisk-Health).

  4. Выполните команды ниже: ```bash python3.13 --version cd ~/yaDisk-Health python3.13 -m venv .venv .venv/bin/python -m pip install requests ``` 5. Проверьте оконный компонент: ```bash .venv/bin/python -m tkinter ``` Закройте тестовое окно и запустите программу: ```bash .venv/bin/python yaDiskHealth.py ``` ### 6. Получите OAuth-токен Яндекса OAuth-токен — ключ, который разрешает программе обращаться к вашему Диску. Минимально требуется право чтения **`cloud_api:disk.read`**.

  Успешный результат означает, что начало файла удалось получить во время проверки. Он **не подтверждает целостность всего файла**. Сетевой сбой также может дать ошибку проверки. <img width="1150" height="718" alt="image" src="https://github.com/user-attachments/assets/d60a48be-e345-4295-bb19-76b0968a4cf1" /> <img width="1755" height="718" alt="image" src="https://github.com/user-attachments/assets/e1ba415b-e30a-4b78-8325-aea63a4a5b52" /> <img width="1372" height="718" alt="image" src="https://github.com/user-attachments/assets/736b855b-dd04-4e51-99b4-6697ba567478" /> ## Возможности

  查看 README 缓存打开 GitHub
• 0xS4N4TG/kinetographTypeScript · 335 Stars

  视频渲染

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  A blazingly fast programmatic video engine, made for agents.

  编报时的 AI 中文解读

  Kinetograph 是一个面向代理（agents）的快速程序化视频引擎，支持用 TypeScript、HTML 或 HyperFrames 项目编写视频，并通过 Rust GPU 合成器渲染。适用于需要自动化生成视频的开发者或 AI 代理。

  编报时的 README 依据

  determinism shim), `@kinetographdev/audio`, the `kinetograph` CLI, and `@kinetographdev/native` (the addon, `crates/node`). ```sh npm ci npm run build # TS packages and a debug addon npm run build:release -w @kinetographdev/native # the addon for timing npx kinetograph still examples/rajz-cards/index.tsx 60 -o out/rajz-60.png npx kinetograph strip examples/rajz-cards/index.tsx -n 12 -o out/rajz-strip.png npx kinetograph render examples/rajz-cards/index.tsx -o out/rajz-cards.mp4 npx kinetograph inspect examples/rajz-cards/index.tsx --frame 60 --json ```

  A project imports from `remotion` and `@remotion/*` as it would for Remotion (the CLI maps them onto `@kinetographdev/core`) and from `@kinetographdev/core` for the native elements and effects. Every command takes `--json`; `npx kinetograph mcp` serves the same commands over MCP; [Use with agents](packages/site/content/docs/guides/agents.mdx) has the client setup and the agent skill. `npm test` runs the CPU tests; the GPU ones run with `KINETOGRAPH_GPU_TESTS=1`. ## HTML pages and HyperFrames projects

  is ported natively, and Chrome for Testing is used only by tests, as a reference to compare against. Node hosts the engine so every npm package and Web API is there; the engine is a native addon so pixels never cross into JavaScript. Existing crates do text shaping, layout, CSS and rasterising; Kinetograph writes the glue. ## Projects in TSX Node 22 or later. The npm workspaces are `@kinetographdev/scene` (scene types and effect helpers), `@kinetographdev/react` (the reconciler and native elements), `@kinetographdev/core` (Remotion's API and packages, the frame loop, the

  loads from the web first and pins it in `kinetograph.lock.json`; `--offline` never downloads. Clips (`data-start`, `data-duration`), nested compositions (`data-composition-src`), `<video>` and `<audio>`, `data-fx-chain` audio effects and `data-color-grading` work as in HyperFrames: the details are under "HyperFrames projects" on the kinetograph site's compatibility page (`/docs/compat/#hyperframes-projects`). ## Build from source Use the rustup toolchain (stable, edition 2024) and a GPU that wgpu supports (Metal on macOS); no display is needed.

  `render` links FFmpeg, so building needs its development libraries and `pkg-config` to find them (`brew install ffmpeg pkgconf` on macOS). Without `pkg-config`, set `FFMPEG_DIR` to the FFmpeg prefix instead (`FFMPEG_DIR=$(brew --prefix ffmpeg)`). ```sh cargo build --release ./target/release/kinetograph render examples/papercut-card.json -o out/card.mp4 ``` How a frame is made, the code map, the Rust CLI, the layer cache and frame reuse are in [packages/cli/skills/kinetograph/references/INTERNALS.md](packages/cli/skills/kinetograph/references/INTERNALS.md). ## License

  - **Inspect and render:** props, the scene tree, each layer's resolved fields and renders, all in the same window. ## Status **Beta, version 0.2.5.** It works and renders real projects, but APIs can still change between versions and some features are missing. The core renderer, effects, authoring layer, studio, Remotion compatibility on a native DOM engine (CSS, SVG, Canvas 2D, WebGL fragment shaders) and an own 3D layer are built. The papercut test project renders all 988 of its frames natively. What is left is listed under known issues on the kinetograph site (`/docs/known-issues/`).

  查看 README 缓存打开 GitHub
• Humpalumps/Ultra-WaterJavaScript · 304 Stars

  海洋渲染、实时模拟、WebGL演示

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  Real-time FFT ocean renderer for three.js (WebGL2): shore breakers, foam, caustics, underwater rendering and buoyancy at 200+ fps

  编报时的 AI 中文解读

  这是一个用于 three.js 的实时海洋渲染器，基于 WebGL2，包含波浪、泡沫、焦散、水下效果和浮力模拟，并附带一个演示岛屿场景，适合用于游戏或可视化项目中的逼真海洋效果。

  编报时的 README 依据

  # Ultra Water Real-time ocean renderer for three.js (WebGL2), with a demo island scene. ```bash npm install npm run dev ``` Open the printed localhost URL. Orbit with the mouse. Pick shots from the `camera` dropdown (they also set the sun). The `benchmark (uncapped)` button renders 300 frames back-to-back with the render loop paused and times them to GPU completion with a fence. It also runs once, 3 s after load. ## Performance

  Measured with the in-page benchmark on an RTX 3060 (Chrome, ANGLE/D3D11), 4× MSAA, all features on, at 1954×1692 (3.3 MP, the pixel count of 1440p; `devicePixelRatio` 2 on the browser window): | Shot | ms / frame | fps (uncapped) | |---|---|---| | Open ocean | 3.30 | ~303 | | Breakers | 4.17 | ~240 | | Golden hour | 4.20 | ~238 | | Lagoon | 4.48 | ~223 | | Flotsam (10 floating props) | 4.62 | ~217 | | Underwater | 4.71 | ~212 | | Aerial | 4.80 | ~208 |

  Camera presets `Flotsam` and `Underwater` spawn a field of floating props. The `Floaters & underwater` folder has spawn, drop and clear buttons, plus drag, buoyancy, scattering, shaft and meniscus controls. ## Not in yet - Wakes behind moving objects. Floaters only disturb the water when they drop in (rings and splash). - Floaters don't collide with each other. - Breakers are heightfield waves, so they lean but never curl over into a tube. - The surface mesh is a fan inside 1 m of the camera, so detail right at the lens is coarse when sitting on the waterline.

  查看 README 缓存打开 GitHub
• brainstormity/LightPostPython · 276 Stars

  社交媒体调度、多账号发布、定时发布

  Stars 统计更新于 2026/10/30 07:40:45

  查看原始资料

  作者描述

  A lightweight system to automate posting to all of your socials from one dashboard

  编报时的 AI 中文解读

  LightPost 是一个轻量级、自托管的社交媒体调度工具，支持从单一仪表板连接多个账户、撰写帖子并立即发布或定时发布。适用于需要管理多个社交平台账号的个人或小团队。

  编报时的 README 依据

  # LightPost **A lightweight system to automate posting to all of your socials from one dashboard.** A lightweight, self-hosted social media scheduler. Connect your accounts, write a post, and publish now or schedule it for later—all from one dashboard.  ## Features - **Multi-account publishing:** connect multiple accounts, including several on the same platform, and post to them together.

  Requires **Python 3.11+ on Linux or macOS**. ```sh git clone https://github.com/brainstormity/LightPost.git cd LightPost python3 -m venv .venv .venv/bin/python -m pip install -r requirements.txt .venv/bin/python -m lightpost ``` Open [localhost:3000/app](http://localhost:3000/app), connect an account, and create your first post. New workspaces start empty. Open **Setup Guide** in the sidebar for platform-specific setup.

  Images are limited to **1 MB each**. Use **Add Image** to import a direct public HTTPS JPEG URL. Keep its URL available until delivery completes for Instagram, Threads, and Discord; your server can remain private. Platform-specific limits appear in the composer and Setup Guide. Videos are not supported. Bring your own authorized platform credentials. API fees, token expiry, permissions, and account eligibility still apply. Initial OAuth consent flows are not included; connect using credentials you obtain from each platform. ## Run locally or on your server

  | TikTok | _Under development_ | | Pinterest | _Under development_ | | Google Business Profile | _Under development_ |

  查看 README 缓存打开 GitHub
• BryceYuuu/RUNPython · 238 Stars

  移民路线参考、政策变化追踪、生活成本对比

  Stars 统计更新于 2026/10/30 07:40:48

  查看原始资料

  作者描述

  RUN · 润：海外移居手册（欧洲 36 国 · 110 条路线 · 主要移民地区）

  编报时的 AI 中文解读

  这是一本面向海外移居者的手册，涵盖欧洲36国及全球主要移民地区的110条移民路线，提供数字游民、技术移民、创业自雇、投资居留等各类路线的关键数字、流程和材料清单，并附有2025-2026年政策变化提醒。适合计划移居海外、需要了解各国移民政策和路线对比的人群。

  编报时的 README 依据

  **[🧭 我该走哪条路](docs/routes/choose-route.md)** ｜ **[🌏 主要移民地区](docs/global/README.md)** ｜ **[🇪🇺 欧洲 36 国](docs/europe/README.md)** ｜ **[📊 路线对比](docs/europe/README.md#compare)** ｜ **[📑 文件准备](docs/practical/china-documents.md)** </div> > [!CAUTION] > 本手册仅供信息参考，**不构成**移民、法律、税务或投资建议。2025–2026 年大量路线关闭或改规则，网上很多旧攻略已经失效——任何决定前请以各国官方页面为准。详见 [免责声明](DISCLAIMER.md)。 --- ## 🚪 你是哪种情况？ <table> <tr> <td width="33%" valign="top" align="center"> ### 💻 **[数字游民](docs/routes/digital-nomad.md)** 为境外雇主或客户远程工作，想换个国家生活 _{15 条欧洲路线 · 月入 €330 起} </td> <td width="33%" valign="top" align="center">

  <td align="center"><a href="docs/europe/united-kingdom.md">🇬🇧<br><b>英国</b></a><br>_见欧洲篇</td> </tr> </table> --- ## 🇪🇺 欧洲 36 国 > [!NOTE] > 每个国家页都有 **🏙️ 生活与身份档案**（生活成本、房租、安全、平均工资、个税、护照、永居和入籍年限、双重国籍、中国护照入境方式），以及每条路线的关键数字、流程图和材料清单。横向对比见 [生活与身份对比表](docs/europe/README.md#profiles)。 _{💻 数字游民 · 🧑‍💻 技术工作 · 🚀 创业 · 🧾 自雇 · 🏖️ 被动收入 · 💰 投资 · 🛂 入籍 · 🧩 其他} <!-- AUTO:countries --> <table>

  | 🔗 | [资源导航](docs/resources.md) | 官方门户、类似网站、其他润学项目 | | 📊 | [programmes.csv](data/programmes.csv) | 99 条路线的结构化数据，可直接用 Excel 打开 | --- ## 🧭 一张图选路线 ```mermaid flowchart LR A([你的收入从哪来？]) --> B(目的国雇主 offer) A --> C(境外雇主/客户) A --> D(自己当老板) A --> E(一笔长期闲置资金) B --> B1[🧑‍💻 技术工作 / 欧盟蓝卡<br>德国 · 荷兰 · 爱尔兰 · 法国] C --> C1[💻 数字游民签证<br>西班牙 · 葡萄牙 · 希腊 · 克罗地亚] D --> D1[🚀 创业 / 自雇<br>爱尔兰 STEP · 芬兰 · 荷兰] E --> E1[💰 投资居留<br>希腊 · 塞浦路斯 · 马耳他 · 匈牙利] ``` _{更细的判断见 [我该走哪条路？](docs/routes/choose-route.md)} --- <details> <summary><b>📁 项目结构 & 如何参与</b></summary> ```text

  <div align="center"> # RUN · 润 ### 一本海外移居手册 数字游民 · 技术移民 · 欧盟蓝卡 · 创业自雇 · 投资居留       [](README.en.md)

  查看 README 缓存打开 GitHub
• anomalyco/hokumRust · 189 Stars

  通知管理、主题定制

  查看原始资料

  作者描述

  Notifications for Wayland that look like they belong on a computer from this decade. Drawn with Rune.

  编报时的 AI 中文解读

  Hokum 是一个为 Wayland 合成器设计的通知守护进程，提供现代化外观的通知卡片，并支持主题定制、应用级通知管理和设置界面。适合使用 niri、sway、Hyprland 等支持 layer-shell 的 Wayland 用户。

  查看 README 缓存打开 GitHub
• paipiange/nvzhuangHTML · 167 Stars

  服装参考图浏览、AI换装试穿

  查看原始资料

  作者描述

  1407 张 AI 女装参考图集与虚拟试穿网站，支持中英检索、OpenAI Images / Gemini / 豆包及自定义 Base URL、API Key、模型。https://github.com/paipiange/nvzhuang

  编报时的 AI 中文解读

  这是一个收录1407张AI生成女装参考图的网站，支持中英文检索、分类筛选和在线浏览，并可选接入AI换装后端实现虚拟试穿。适合服装设计、电商广告等需要服装参考素材的用户。

  查看 README 缓存打开 GitHub
• andyhuo520/vibe-knowledgeJavaScript · 162 Stars

  视频生成、知识科普

  查看原始资料

  作者描述

  Vibe 知识大赏：一个知识点 → 一条有旁白、每句有出处的精品知识短片（Claude Code skill · 44 套风格 · 代码逐帧成片）

  编报时的 AI 中文解读

  这是一个Claude Code skill，用于将一个知识点制作成有旁白、每句有出处的精品知识短片，全程在对话中完成选题、查证、写稿、配音、代码成片、质检和发布包。适合使用Claude Code并希望自动化生成知识类短视频的用户。

  查看 README 缓存打开 GitHub
• 10291124jcjc-dotcom/ai-anime-seriesPython · 160 Stars

  AI动画生成、提示词生成、视频剪辑辅助

  查看原始资料

  作者描述

  AI 动画单元剧：固定原创角色，一句话出一集。角色设定图 + 宫格故事板 + 即梦 Seedance 一次生成。Claude Skill.

  编报时的 AI 中文解读

  这是一个用于制作AI动画单元剧的完整流程和工具集，包含Claude Skill和辅助脚本。用户固定几个原创角色后，每集只需一句话即可生成剧本、提示词和故事板，并配合即梦画布生成视频。

  查看 README 缓存打开 GitHub
• cutetailbud/ff-ruyijsC++ · 160 Stars

  网页自动化、JS逆向执行、网络请求模拟

  查看原始资料

  作者描述

  Run any site JavaScript from Python with the real Firefox engine. Headless SpiderMonkey + Necko for automation and scraping.

  编报时的 AI 中文解读

  ruyiJS 是一个从 Python 运行网站 JavaScript 的库，使用 Firefox 的真实引擎（SpiderMonkey）和网络栈（Necko），无窗口、无渲染器，专为网页抓取、JS 逆向、数据收集和绕过机器人检测而设计。适合需要执行复杂 JavaScript 或处理反爬挑战的开发者。

  查看 README 缓存打开 GitHub
• jmanico/grim-prompter未标注语言 · 155 Stars

  威胁建模、安全审查

  查看原始资料

  作者描述

  Threat modeling for AI coding agents: point Claude Code, Codex or any agent at your repo and get an evidence-traced threat model. 36 architectures, 29 methods, MIT.

  编报时的 AI 中文解读

  Grim Prompter 为 AI 编程代理提供威胁建模提示词和技能，可对代码仓库进行结构化安全审查，生成带证据的威胁模型报告。适用于使用 Claude Code、Codex 等代理进行安全评估的开发者。

  查看 README 缓存打开 GitHub
• lovisdotio/video-to-gameHTML · 111 Stars

  视频转游戏、游戏开发、资产生成

  查看原始资料

  编报时的 AI 中文解读

  这是一个教编码代理将游戏视频转化为可玩浏览器游戏的技能，适用于Claude Code、Codex、Cursor等代理，通过视频模型设计游戏外观，代理逐帧分析并生成资产和代码。

  查看 README 缓存打开 GitHub
• floatboatai/OpenMinutesTypeScript · 106 Stars

  录音转写、AI归类、全文搜索

  查看原始资料

  作者描述

  OpenMinutes · 录音笔工作台：MemoTalk S1 录音的导入 · 转录 · AI 归类 · 检索复盘（Floatboat Workbench 开源实现）

  编报时的 AI 中文解读

  OpenMinutes 是一个将 MemoTalk S1 录音笔的 MP3 文件转换为可搜索、可分类、可复盘的文本资产的工作台应用，适用于需要管理大量录音并从中提取信息的用户。

  查看 README 缓存打开 GitHub
• ScrewPantherRampart/clean-v4.0.3未标注语言 · 102 Stars

  系统优化、隐私保护

  查看原始资料

  作者描述

  Published: 2026-10-28

  编报时的 AI 中文解读

  Clean 是一个用于 Windows 10/11 的 PowerShell 7 脚本，通过可逆的清单移除预装应用、禁用遥测并调整服务，以优化系统。

  查看 README 缓存打开 GitHub
• Novavyalimit/vanta-v2.1.0未标注语言 · 102 Stars

  游戏增强

  查看原始资料

  作者描述

  Published: 2026-10-28

  编报时的 AI 中文解读

  Vanta 是一款面向 Minecraft 玩家的 Fabric 实用客户端，提供 120 多个模块和 ClickGUI，支持战斗、移动、视觉增强等功能，适用于 Java 21 环境。

  查看 README 缓存打开 GitHub
• ShadyHero/Eden-NS2C++ · 97 Stars

  游戏模拟

  查看原始资料

  作者描述

  please refer to https://github.com/0x2593/Eden-NS2 for updates and detailed instructions

  编报时的 AI 中文解读

  Eden 是一款免费开源的 Switch 1 模拟器，用 C++ 编写，支持 Windows、Linux、macOS、Android、FreeBSD 等多个平台，旨在运行商业游戏。

  查看 README 缓存打开 GitHub
• w0ldan/exoprocTypeScript · 78 Stars

  进程检查、内存操作、函数注入

  查看原始资料

  作者描述

  TypeScript toolkit for Windows x64 native systems programming, memory manipulation, and process instrumentation.

  编报时的 AI 中文解读

  Exoproc 是一个用于 Windows x64 原生系统编程的 TypeScript 工具包，提供统一的命令行界面和 MCP 服务器，支持进程检查、内存操作和函数注入等逆向工程与调试功能。它适用于安全研究人员、逆向工程师以及需要深入了解或操控运行中进程的开发者。

  查看 README 缓存打开 GitHub
• Xpropel/Pi-RemoteTypeScript · 77 Stars

  远程控制、会话镜像、多设备同步

  查看原始资料

  作者描述

  Mirror and remotely control pi terminal sessions from any browser — official pi extension + lightweight relay + unified desktop/mobile UI.

  编报时的 AI 中文解读

  Pi-Remote 是 pi 终端编码代理的配套工具，通过浏览器实时镜像和控制 pi 会话，支持远程发送消息、执行命令、切换模型等，适用于需要远程管理 pi 的用户。

  查看 README 缓存打开 GitHub
• fatihfurkan46/SignalWeavePython · 72 Stars

  数据规范化、指标聚合、数据报告

  查看原始资料

  作者描述

  Signal weaving toolkit - combine noisy service metrics into stable, explainable health signals

  编报时的 AI 中文解读

  SignalWeave 是一个用于处理异构服务遥测数据的命令行工具，可将来自不同 AI 框架的事件规范化并聚合成稳定性、质量和成本等指标。适合需要统一监控多个 AI 服务运行状况的开发者或运维人员。

  查看 README 缓存打开 GitHub
• zklmiao/CodeWarGo · 68 Stars

  编程对战

  查看原始资料

  作者描述

  网页端 Python 编程对抗游戏：解题速度即战斗力 · A turn-based competitive Python programming game

  编报时的 AI 中文解读

  网页端双人Python编程对战游戏，通过解题速度决定战斗效果。玩家在5×5竞技场控制角色，完成题目以攻击、移动、防御或释放技能。适合编程学习者以趣味方式练习解题能力，也适合朋友间对战。

  查看 README 缓存打开 GitHub
• CokePokes/LetraInstaller未标注语言 · 67 Stars

  IPA签名安装、插件注入、应用重签

  查看原始资料

  作者描述

  Modify and install ipa files on the fly.

  编报时的 AI 中文解读

  Letra 是一款在 iOS 设备上直接签名和安装 IPA 文件的工具，支持注入插件、修改应用元数据，并可通过 TrollStore 或 LocalDevVPN 安装，无需电脑。

  查看 README 缓存打开 GitHub
• Huy-Quang-Dao/instantTrackPython · 53 Stars

  运动跟踪

  查看原始资料

  作者描述

  Whole-body motion tracking for the Unitree G1 in 15 minutes on one GPU (SuperMimic-style world model + InstantMimic-style GPU loop, mjlab)

  编报时的 AI 中文解读

  InstantTrack 是一个用于 Unitree G1 机器人全身运动跟踪的训练系统，基于 SuperMimic 和 InstantMimic 方法，在单张 GPU 上 15 分钟内完成训练，适用于机器人运动控制研究。

  查看 README 缓存打开 GitHub
• leisure271/maodie-purge未标注语言 · 53 Stars

  提示词注入、渗透测试

  查看原始资料

  编报时的 AI 中文解读

  Maodie Armor 是一个面向 DeepSeek Harness 的本地提示词工程插件，提供一键注入/卸载、技能包和 AI 渗透控制台，用于稳定注入系统提示词、实时观测拒绝行为并支持配置回滚。

  查看 README 缓存打开 GitHub

查看本期全部收录项目