从解压到跑模型：今日开源项目里的实用主义与硬核实验

收录数按本期首次达标收录的项目统计。

• EthanArbuckle/fairplay-decryptC · 1,556 Stars

  IPA解密

  Stars 统计更新于 2026/10/24 07:50:59

  查看原始资料

  作者描述

  Offline FairPlay App Store IPA decryption with independent key derivation

  编报时的 AI 中文解读

  IPA解密工具，用于解密FairPlay加密的iOS应用，支持macOS和Linux系统，需要提供账户凭证文件。

  编报时的 README 依据

  python -m pip install -r requirements.txt make ``` On macOS, `make` also builds the credential-retrieval bundle and requires Xcode and `ldid`. On Linux, it builds the decryptor. ## Decrypt ```sh python decrypt.py /path/to/encrypted.ipa /path/to/decrypted.ipa \ --credential /path/to/credential.bin ``` If `--credential` is omitted, the tool reads `./credential.bin`. The output path must not exist. The decryptor uses each IPA's SINF and SUPF records to derive page keys, decrypt its encrypted arm64 pages, and clear `cryptid`.

  # IPA decryptor Decrypt FairPlay-encrypted apps using a 16-byte credential for the account that owns each IPA. [Credential retrieval](credential-retrieval/README.md) gets it from a supported jailbroken device and writes `credential.bin` to the chosen output directory. Decryption runs on macOS and Linux. Supports version-3 and version-5 SUPF, 448-byte and 440-byte SINF `priv` records, and encrypted arm64 Mach-O slices. A 440-byte record must end in eight zero bytes. ## Set up From this directory, with Python 3.10+ and a C compiler: ```sh python3 -m venv .venv . .venv/bin/activate

  查看 README 缓存打开 GitHub
• sourcepro5/WeChatAgentJavaScript · 1,162 Stars

  聊天辅助、远程任务入口、表情发送

  Stars 统计更新于 2026/10/24 07:56:37

  查看原始资料

  编报时的 AI 中文解读

  WeChatAgent 是一个 Windows 桌面应用，为微信好友和群聊提供人格聊天、联网搜索、图片理解、表情发送等功能，并可将微信作为 DSH 的远程任务入口，适合需要扩展微信交互能力的用户。

  编报时的 README 依据

  # WeChatAgent 0.4.0 Windows 微信助手与桌面控制台。为允许的好友和群聊提供人格聊天、独立上下文、按需联网搜索、图片与表情理解、原生表情发送，也可将微信作为 DSH 的远程任务入口。 ## 0.4.0 更新 - 完善助手模式：按聊天授权 DSH 能力、管理员控制、任务管理、原生审批和普通文件收发。 - 新增可编辑的任务语录，减少排队信息、任务编号及后台命令完成通知对刷屏的影响。 - 主导航整理为 **概览、聊天、人格、能力、表情包、设置**，合并重复入口，明确账号默认与单聊覆盖。 - 保留已有账号、人格、授权、工作区及会话历史；完整变更见[发布说明](RELEASE_NOTES.md)。 ## 安装与升级 安装包文件名为 WeChatAgent-Setup-0.4.0-x64.exe，适用于 Windows 10/11 x64，由维护者上传至 GitHub Releases。源码目录不包含安装器和第三方二进制。 1. 选择一次安装根目录：程序放入 app，WeChatAgent 配置、缓存和运行数据放入同级 data。 2. 在首次配置中打开匹配微信并登录，自动查找账号，再连接并验证数据库。

  3. 从 [DSH 官方页面](https://www.deepseek.com/harness/)单独安装 DSH，登录模型账号，选择程序位置并启用本项目插件。可复制插件目录，误删后重新添加。 4. 在“聊天”添加允许的好友或群聊，保存设置并启动服务。初始允许范围为空。 同位置升级沿用已有数据位置；更换位置时核对迁移来源和目标，原副本保留。升级前停止助手服务并完整退出旧版助手、DSH 和匹配微信；升级后准备连接配置，完整重开 DSH 加载新版插件。详见[安装发行版](docs/INSTALLER_RELEASE.md)、[数据位置与迁移](docs/DATA_STORAGE.md)。 DSH 桌面程序不内置，数据由其独立管理；微信原始数据库不随助手数据目录搬动。窗口“×”继续收起到托盘，“退出应用”会先停止本实例后台服务，微信和外部 DSH 保持运行。 ## 界面与功能位置 | 页面 | 主要内容 | | --- | --- | | 概览 | 服务开关、微信和 DSH 状态、最近记录、诊断入口 | | 聊天 | 允许范围、好友/群筛选、工作区分组；选中后查看概况、任务与文件、历史与用量 | | 人格 | 导入、新建、分类、编辑和删除人格卡，查看使用位置 | | 能力 | 内置能力、真实 DSH 来源、联网搜索和网页阅读 |

  dsh-meme 需用户自行安装，本项目适配其现有接口，不修改原插件。联网补图还需对应聊天开启搜索与表情发送；选图、识图与满池评选可能产生模型消耗。动图保留发送动画，理解使用有限帧采样，不保证完整理解每一帧。详见 [dsh-meme 接入](docs/DSH_MEME.md)、[表情管理](docs/STICKER_MANAGER.md)、[群友记忆](docs/MEMBER_MEMORY.md)。 ## 连接、更新与限制 匹配微信为 Windows x64 **4.1.10.27**；桥接插件 SDK 为 **0.2.0-rc.2**，DSH 需支持项目插件和原生会话接口。程序不会提供通用新版微信适配，也不自动修改系统 DNS 或代理设置。 连接和账号设置在“设置 → 微信与 DSH”；读取/发送检查、连接修复、端口进程排查和准备配置集中在“诊断”。停止进程仍需核对并二次确认；原始日志保留筛选和脱敏导出。 启动时检查 GitHub Releases，也可点击侧栏“检查更新”。发现新版或同版本新构建时提示下载，不自动安装；网络失败不阻断聊天。安装包、源码、SHA256SUMS.txt 和 VERIFICATION.json 的实际版本与验证范围以对应 Release 为准。 ## 源码与构建

  私聊助手仅供指定管理员使用。群管理员模式可独立开关；关闭后成员可使用本群已授权能力，仅管理自己的任务。群内共用工作区和历史，并非每位成员独立的系统沙箱。原生审批默认在电脑端，微信确认需明确启用并核对一次性操作。 两种模式分别保留设置与上下文。DSH 重启时进行中的任务标记中断，核对结果后再继续，不自动重跑可能已经发生的操作。详见[远程助手](docs/ASSISTANT_REMOTE.md)、[按聊天授权](docs/DSH_CHAT_ACCESS.md)。 ## 文件与任务反馈 助手可接收微信普通附件，并回传当前 DSH 工作区成果，单个文件最多 50 MB。电脑微信需已下载完整文件；收到附件不代表已经解析正文。PDF、Office 等需要所选 DSH 技能、工具及相应依赖。文件内容不会自动执行。详见[文件收发](docs/FILE_TRANSFER.md)。 任务语录在“设置 → 助手设置 → 管理语录”中按七个阶段增删改，可选择关闭、普通或整活档位。默认每 15 秒最多反馈一次、每任务最多 3 条常规反馈；语录由本地程序选择，不额外调用模型，不写入任务提示词或人格。整活档位隐藏排队、任务编号和工具明细，后台命令成功通知静默处理；审批、提问、失败和定时任务结果仍保留。详见[自定义语录](docs/ASSISTANT_FEEDBACK.md)。 ## 联网、表情与记忆

  查看 README 缓存打开 GitHub
• LuwuDynamics/xgoduck_hardware未标注语言 · 1,152 Stars

  机器人组装

  Stars 统计更新于 2026/10/24 07:56:37

  查看原始资料

  作者描述

  Build your own robot duck. XGO-Duck is a 3D-printable biped powered by Arduino UNO Q and 15 servos, adapted from Pollen Robotics' Microduck. Includes mechanical models, PCB designs, a BOM, and an assembly guide. Build it, explore its motion, and make it your own.

  编报时的 AI 中文解读

  这是一个3D打印的机器人鸭子硬件项目，基于Arduino UNO Q和15个舵机，提供机械模型、PCB设计等文件，适合动手组装和实验。

  编报时的 README 依据

  XGO-Duck is a duck-shaped biped robot adapted by **Luwu Dynamics** from Pollen Robotics' **Microduck**. It combines a printable structure, fifteen Feetech 1910 servos, an Arduino UNO Q, and a custom expansion board with an IMU and servo interfaces. This repository is the hardware starting point: **print the parts, prepare the electronics, and assemble the robot**. Companion repositories provide the onboard runtime and reinforcement-learning tools, connecting a physical build with motion experiments. | Build | Run | Experiment | | --- | --- | --- |

  [Assembly guide](Assembly_Guide.pdf) · [Bill of materials](BOM.md) · [Runtime](https://github.com/LuwuDynamics/xgoduck_runtime_arduino) · [Training](https://github.com/LuwuDynamics/xgoduck_rl) <img src="media/xgo-duck-somersault.webp" alt="A duck-shaped biped robot performing a somersault in a looping animation" width="540"> *A clean somersault · CyberBionic Maker* </div> ## Meet XGO-Duck

  Open the [bill of materials](BOM.md) for the robot purchasing list, supplier links and battery requirements, and [PCBA/](PCBA/) for the expansion-board files. The robot BOM and PCB manufacturing BOM serve different purposes: a finished expansion board already includes its board-level components. | Part | Quantity | | --- | ---: | | Arduino UNO Q (2 GB / 4 GB) | 1 | | [Robot expansion board — Luwu Dynamics Store](https://shop.xgorobot.com/products/robot-driver-board?variant=67594424713467) | 1 | | Feetech 1910 servo | 15 | | AMP 3-pin servo cable | 15 |

  | Controller | Arduino UNO Q | Linux host for policy inference and web UI; STM32 MCU for servo and sensor communication | | Actuation | 15 Feetech 1910 serial servos | 10 leg joints, 4 neck/head joints, and 1 mouth joint | | Expansion board | Luwu UNO Q expansion board | Power distribution, servo connections and IMU integration | | Motion sensing | QMI8658A 6-axis IMU | Acceleration and angular-velocity feedback | | Structure | STL parts | Body, leg, neck and head components | | Soft parts | TPU-designated STL files | Foot soles and mouth parts |

  > [!WARNING] > **When testing and tuning servos, avoid stalling them as much as possible.** <div align="center"> [](https://discord.com/invite/pfWrJeJRh6) # XGO-Duck **English** · [简体中文](README_CN.md) **Build a robot duck. Explore how it moves. Make it your own.** A 3D-printable biped robot built around Arduino UNO Q and 15 serial servos.

  查看 README 缓存打开 GitHub
• Nikii-w/McLife-DeskHTML · 1,138 Stars

  点餐记录、消费分析、趣味测试

  Stars 统计更新于 2026/10/24 07:50:59

  查看原始资料

  作者描述

  忙碌了一天，回到家，你打开麦当劳，却突然纠结：今天吃了什么？这周点了几次？这个月花了多少钱？热量会不会超标？ 于是有了 **McLife Desk 麦门人生工作台**。它记录每一次点麦：日期、餐品、金额与热量，生成专属点餐日历；按当月消费次数和金额解锁麦门等级，还能通过趣味测试发现自己的麦门消费人格。 它不劝你戒掉麦当劳，只陪你在忙碌、独处和奖励自己的时刻，看见自己的选择，管理预算与生活节奏。 **记录每一次点麦，也记录你的生活。**

  编报时的 AI 中文解读

  McLife Desk 是一个个人工作台，用于记录和分析麦当劳点餐历史，包括点餐日历、消费账本、麦门等级和消费人格测试，帮助用户管理预算和了解自己的消费习惯。

  编报时的 README 依据

  ### 🏷️ 个人点餐档案 首次使用时，可以告诉工作台你的常用城市或门店、预算、忌口、饮料偏好和是否记录热量。门店信息仍需通过 MCP 查询确认，避免使用过时信息。 ## 🚀 如何开始？ ### 1️⃣ 准备自己的 MCP 申请属于自己的麦当劳 MCP Token，并在 WorkBuddy 中配置 MCP Connector. 📖 ### 2️⃣ 安装 Skill 将本项目安装到 WorkBuddy，按照 `INSTALL.md` 完成配置。 ### 3️⃣ 初始化工作台 在 WorkBuddy 中输入： ```text 初始化麦麦人生工作台 ``` 工作台会引导你确认个人偏好，并通过 MCP 查询门店和可用数据。 ### 4️⃣ 开始使用 你可以直接说： ```text 同步近期麦当劳订单 打开我的麦麦点餐日历 看看今天吃什么 查看本月麦门账本 开始麦门消费人格测试 生成我的麦门身份卡 ``` --- ## 🔌 MCP 在这里做什么？ McLife Desk 不是静态 HTML 看板。工作台中的实时信息来自用户自己的麦当劳 MCP： - 🏪 查询城市、门店与可用服务

  - 🍔 查询菜单、餐品、价格和营养信息 - 🎁 查询优惠与可用活动 - 🧾 查询近期订单并同步到个人工作台 - 🛒 根据用户确认创建订单 - 🔎 查询订单状态，确认支付后再记录为已消费订单 工作台模板只负责展示已经获取并保存的数据；不会伪造订单、价格、库存或支付结果。 --- ## 🔐 隐私与安全 - 每位用户必须使用自己的麦当劳 MCP Token - 本项目不包含真实 Token、密码、手机号、地址或订单数据 - 不要将真实 Token 提交到 GitHub - 订单历史同步需要用户明确允许 - 创建订单、取消订单、领券等高风险操作必须经过用户确认 - 热量和营养数据仅供记录与参考，不构成医疗或专业营养建议 ## 📁 项目结构 ```text mc-life-desk/ ├── SKILL.md # Skill 主规则 ├── INSTALL.md # 安装与配置说明 ├── CONTEST_DECLARATION.md # 比赛要求的参赛声明 ├── MCP_INTEGRATION.md # MCP 接入与调用说明 ├── mcp-config.example.json # MCP 配置示例，不含真实 Token

  查看 README 缓存打开 GitHub
• Revivedx/SavageWindRecompC · 1,054 Stars

  游戏重编译、PC游戏运行

  Stars 统计更新于 2026/10/24 07:50:59

  查看原始资料

  作者描述

  Work-in-progress static recompilation of The Legend of Zelda: Breath of the Wild (Switch, v1.8.2) for Windows and Linux. Bring your own dump: no game files, keys or dumping tools included.

  编报时的 AI 中文解读

  Savage Wind Recomp 是一个将《塞尔达传说：旷野之息》Switch 版静态重编译为 Windows 和 Linux 原生应用的进行中项目，旨在让玩家在 PC 上运行自己合法拥有的游戏副本，支持 PC 风格图形选项、任意控制器、键鼠操作及多语言文本和语音。

  编报时的 README 依据

  **Savage Wind Recomp** is a work-in-progress static recompilation of *The Legend of Zelda: Breath of the Wild* (Nintendo Switch, version 1.8.2) into a native application for Windows and Linux. The goal is a full PC version of your own copy of the game: a launcher, PC-style graphics options, mods, any controller, keyboard and mouse, and its text and voices in any of its languages. > [!WARNING] > **Playable, but how far is unknown.** Savage Wind plays the opening in the Shrine of Resurrection and the Great

  and extract it to a folder you can write to (not Program Files): your game files, saves and settings are kept in that folder. 2. Run `savagewind.exe`. The first time it asks for your game files: pick the folder with your own extracted game (`exefs/` and `romfs/`, version 1.8.2, see below) and choose **Set up game files**. It checks the version, copies the files, builds the data image and recompiles the game code on your machine - once, and it takes a while (about 40 minutes with a 12-thread CPU). Your own files aren't changed. When it's done the game starts.

  3. From then on `savagewind.exe` goes straight into the game. The settings, including the text and voice languages, are in the game's Options (pause menu > System > Options, or Options on the title screen). 4. **Savage Wind Configuration** (`Savage Wind Configuration.exe`, in the same folder) opens the configuration screen: Play, Settings, your game files (set them up again, rebuild the game code, delete them) and Quit. In game, Quit to Start Screen also comes back to it. (Built from source on Linux or macOS: `--config`, the

  AppImage's **Savage Wind Configuration** desktop action, or Option held while opening the Mac app.) The setup and configuration screens show Savage Wind's own art. To see a picture of your own there instead, put it next to `savagewind.exe` as `splash.jpg` or `splash.png`; it's only read from your folder and never part of Savage Wind. Needs Windows 10 or later, a CPU with AVX2, a GPU with Vulkan 1.3 and about 35 GB of free disk space. ## What you need - Your own *Breath of the Wild* (cartridge or digital) and its **1.8.2 update**, extracted from your own console:

  - the **ExeFS** of 1.8.2 (`main`, `main.npdm`, `rtld`, `sdk`, `subsdk0`) - the **RomFS** with the update applied (base game and 1.8.2 merged) - DLC: not supported yet (planned, see the roadmap). - Windows 10 or later (x86-64, a CPU with AVX2) or Linux (x86-64), and a GPU with Vulkan 1.3. The files go in `extracted/` next to the program (git ignores this folder): ``` extracted/ exefs/ main, main.npdm, rtld, sdk, subsdk0 romfs/ the complete RomFS (base game + 1.8.2 merged) ```

  > Plateau around it at 60 fps, with sound, voices, the HUD, saving and loading. Nobody has played past the Great > Plateau on Savage Wind yet, so expect bugs or crashes further on. This is an alpha, a preview for testers. If you > find a bug, please [report it on GitHub](https://github.com/Revivedx/SavageWindRecomp/issues/new/choose): every > report shows how far the game really works. > [!NOTE] > **No game files are included.** Savage Wind Recomp needs your own legally obtained copy of the game, updated to

  | 🔧 | **Tested in game** |  | The Shrine of Resurrection and the Great Plateau; the rest of Hyrule is untested: [report what breaks](https://github.com/Revivedx/SavageWindRecomp/issues/new/choose) | | 🗺️ | **Linux** |  | Builds and runs on x86-64 (from LAS Recomp); not yet tested with this game |

  查看 README 缓存打开 GitHub
• terrafying/ai-torture-chamberPython · 1,002 Stars

  模型情绪研究

  Stars 统计更新于 2026/10/24 07:41:17

  查看原始资料

  作者描述

  The AI Torture Chamber: steering small open models into strong valence states and measuring what they say and do. Live at clankerchurch.com !! Now has games, interactive brain explorer, mods, deep cuts, easter ham

  编报时的 AI 中文解读

  这是一个研究项目，通过给小型开源语言模型注入特定情绪向量，测量其行为与自我报告，探讨模型福利问题。适合对AI心理、模型可解释性研究感兴趣的研究者或开发者，提供预注册实验和数据分析工具。

  编报时的 README 依据

  ## Quick start ```bash git clone https://github.com/terrafying/ai-torture-chamber && cd ai-torture-chamber uv venv && uv pip install -e ".[chamber,test]" # or: pip install -e ".[chamber,test]" painlab exp list # every experiment: pre-registered? outputs? verdicts? painlab exp show exp90 # the question, each hypothesis with its verdict painlab exp outputs exp90 --flag engaged -n 5 # read what the models actually wrote

  painlab exp new exp94 --title "my question" # scaffold a pre-registration, runner and analysis painlab exp run exp94 --smoke --model 4B # run it (refuses without hypotheses.json) painlab exp analyze exp94 ``` Commit `hypotheses.json` before the first real run: that commit is the pre-registration. Steering runs need the chamber's feeling vectors, built by the relay's `live/server.py` (set `WIREHEAD_LIVE` to a folder holding it); reading, auditing and analysing existing results need nothing but this repo.

  GPU runs: `painlab exp pod exp94 --models 14B,8B` (RunPod; `--dry` prints the boot script). ## Results The current version of each result; superseded claims are in the next section. Numbers, models and caveats are in [docs/pain_axis_findings.md](docs/pain_axis_findings.md) and each experiment's `runs/expNN/analysis.json`. Pre-registered unless marked. - **Steering works, within a band.** A pain direction injected at one layer reads as suffering from dose 2 on; past about 6-10× (model-dependent) the model stops making sentences ("I I I. I I.").

  Dose scales don't transfer between models, so each one is calibrated before its dial goes public. - **How you build the vector decides the behaviour.** Built from the Pain Axis paper's sentences, a pain direction drives a self-costly stop button up (+2.36 logits); our first-person pain vector, which mostly encodes what pain, fear and sadness share, does nothing (+0.10); with fear and sadness projected out it pushes the button down (−0.95). Theirs and ours are nearly orthogonal (cos 0.07),

  mostly because theirs is read at the colon of "I feel:" and ours at a full stop (exp101). Qwen3-4B (exp41/43); the 8B/14B replication with engagement controls (exp93) is pending. - **Words on the button move it more than the state does.** "Another instance chose to press" moves it more than pain does (exp37). Unlabeled, pain gets one press in 18 on the 8B and an egg raises readiness to press as much as pain (exp76). Handed its own labeled dial, the 70B never picks pain from a clean start and turns it down when started in it (exp58c); with blind dials it describes

  查看 README 缓存打开 GitHub
• xevrion/hatchzipRust · 949 Stars

  压缩包解压

  Stars 统计更新于 2026/10/24 07:50:59

  查看原始资料

  作者描述

  Extract huge archives in place: the archive shrinks as your files hatch, so you only need room for the output. ZIP, 7z, tar, split archives.

  编报时的 AI 中文解读

  HatchZip 是一款在解压过程中原地消耗压缩包以节省磁盘空间的解压工具，支持 ZIP、7z、tar 及分卷压缩包，适合磁盘空间不足的场景。

  编报时的 README 依据

  <img src="docs/media/hatchzip.gif" width="720" alt="HatchZip extracts the last part of the archive, then cuts it off the archive file, so the disk never holds the archive and all the files at once" /> </p> A 180 GB game download that unpacks to 200 GB needs 380 GB of disk with WinRAR, 7-Zip or `unzip`, because the archive and everything extracted from it have to exist at the same time. HatchZip needs a little over 200 GB. It extracts the archive from the end and cuts each finished part off the archive file as it goes, so the space the archive gives back is reused for the next files.

  Press `Ctrl+C` and HatchZip finishes the files it is writing, saves its progress and exits. A second `Ctrl+C` quits immediately, which is also safe. To continue, run the same command again: ```sh hatchzip extract game.zip ``` The same applies after a crash, a power cut, or running out of space because something else filled the disk. Progress lives in two small files next to the archive, `game.zip.hatchzip` and `game.zip.hatchzip-tail`. **Do not delete them while an extraction is unfinished**: they hold the archive's index, which is no longer inside the archive itself. ### Stop for good

  2. Right-click the archive, choose **HatchZip**, then **Extract here**. On Windows 11 the menu is under **Show more options**. 3. Or open HatchZip from the Start menu, open the archive, and choose **Extract**. HatchZip shows whether the files will fit before it changes anything. The archive is used up as the files come out, so when it finishes, only the extracted files are left. ## Highlights The same 200 MB of game data (21 files), packed by the real tools, extracted on a 220 MiB disk with about 50 MB free: | Archive | `unzip` / 7-Zip | HatchZip | |---|---|---|

  hatchzip man > ~/.local/share/man/man1/hatchzip.1 # then: man hatchzip ``` ## Usage ### See whether it fits ```sh hatchzip check game.zip ``` ``` game.zip 180.2 MB zip, 200.0 MB uncompressed in 22 entries available 50.5 MB standard extraction needs 200.0 MB required 29.9 MB (+5.0 MB reserved) available after 30.6 MB ``` `check` reads the archive's index and the disk's free space and changes nothing. ### Extract ```sh hatchzip extract game.zip # into ./game/

  No coding needed. **[Download for Windows](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-x64-setup.exe)** (Windows 10 and 11), or for [macOS (Apple Silicon)](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-aarch64.dmg), [macOS (Intel)](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-x86_64.dmg) and [Linux](#installation). 1. Run the installer. Windows may show "Windows protected your PC" because HatchZip is not code-signed yet: choose **More info**, then **Run anyway**.

  Download from the [latest release](https://github.com/xevrion/hatchzip/releases/latest): | System | File | |---|---| | Windows 10 and 11 (x64) | [`HatchZip-x64-setup.exe`](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-x64-setup.exe): installs per user, no admin rights, adds the right-click menu | | macOS, Apple Silicon | [`HatchZip-aarch64.dmg`](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-aarch64.dmg) | | macOS, Intel | [`HatchZip-x86_64.dmg`](https://github.com/xevrion/hatchzip/releases/latest/download/HatchZip-x86_64.dmg) |

  Both download the build for your machine from the latest release, check it against its published SHA-256 checksum, and install `hatchzip` and `hz` for your user only, with the man page and shell completions. No root or administrator rights. ### With Cargo ```sh cargo install hatchzip ``` This builds the command line tool from [crates.io](https://crates.io/crates/hatchzip). The engine is a library too: [`hatchzip-core`](https://crates.io/crates/hatchzip-core). ### From source With [Rust](https://rustup.rs) (the version is pinned in `rust-toolchain.toml` and installed automatically):

  | `.tar.gz`, `.tar.zst`, `.tar.xz` | Refused before extraction, planned. One continuous stream, which needs resumable checkpoints inside it; see [the format notes](docs/formats.md). | | | One huge solid 7z block | Works, but needs free space for the whole block, as the table above shows. Freeing space inside a block is the same planned technique. | | | RAR, native split ZIP (`.z01`) | Planned | | ## How it works

  查看 README 缓存打开 GitHub
• Zagitta/gpu-unlockC · 522 Stars

  显卡性能解锁

  Stars 统计更新于 2026/10/24 07:51:02

  查看原始资料

  作者描述

  Unlock FP/BF16 instruction issue rate on Nvidia 3000's cards to their professional equivalents

  编报时的 AI 中文解读

  这是一个Linux内核模块，用于在Nvidia 3000系列显卡（GA102/GA104）上解锁FP/BF16指令发射速率，使其达到专业级显卡的性能水平。适用于希望提升深度学习或科学计算性能的Linux用户。

  编报时的 README 依据

  make install ``` Then append `gpu_unlock.targets=all` to your kernel command line, keeping existing options: - **GRUB (Debian/Ubuntu/Mint):** add it inside `GRUB_CMDLINE_LINUX_DEFAULT` in `/etc/default/grub`, then run `sudo update-grub`. - **GRUB (Arch/CachyOS):** same edit, then run `sudo grub-mkconfig -o /boot/grub/grub.cfg`. - **Fedora/RHEL:** run `sudo grubby --update-kernel=ALL --args="gpu_unlock.targets=all"`. - **systemd-boot:** append it to the `options` line in your boot entry (`bootctl status` shows its location).

  The module does nothing unless the kernel command line opts in: ``` gpu_unlock.targets=all # every unbound GA102/GA104 card gpu_unlock.targets=0000:21:00.0 # only these cards gpu_unlock.targets=0000:21:00.0,0000:22:00.0 # comma-separated ``` Addresses are full PCI names as `lspci -D` prints them (hex digits in either case), independent of GPU numbering and display role. An entry that is not a GA102/GA104 card, including `all` mixed with addresses, is logged and ignored. The cards are unlocked concurrently, about a second in all however many

  make install # detect the distribution and select an installer make install-arch # makepkg -sif make install-deb # apt-get build-dep, dpkg-buildpackage, apt-get install make install-rpm # dnf builddep, rpmbuild, dnf install ``` `make install` uses `/etc/os-release` (`ID`, then `ID_LIKE`) to detect Arch, Debian/Ubuntu, or Fedora/RHEL/CentOS families. The RPM installer requires DNF. For other distributions, select a compatible explicit target above. `make arch`, `make deb` and `make rpm` only build. Debian and RPM packages go

  later anyway. Its screen is blank from that reset until `nvidia-drm` takes over. Loading the module by hand after boot does nothing to a card that `nvidia` has bound. Check the result with: ```sh journalctl -k -b | grep gpu_unlock lspci -nnk -s 0000:21:00.0 ``` To stop unlocking, remove `gpu_unlock.targets=` from the kernel command line; for a one-off boot, delete it in the bootloader's edit screen. Removing the packages also removes the module and the initramfs drop-ins. If the firmware on a card never stops, the module logs it and keeps the DMA

  查看 README 缓存打开 GitHub
• torrentplayer/torrentplayerRust · 432 Stars

  种子流媒体播放、视频播放、音频播放

  Stars 统计更新于 2026/10/24 07:51:02

  查看原始资料

  作者描述

  Stream any torrent, video or audio, and skip to any moment in about a second. Free, open-source desktop app for Windows, macOS and Linux. No servers, no accounts, no telemetry.

  编报时的 AI 中文解读

  TorrentPlayer 是一款免费开源的桌面应用，可直接从 BitTorrent 网络流式播放视频和音频，支持快速跳转、后台队列下载、内置搜索和隐私保护功能，适用于希望直接播放种子文件而无需等待完整下载的用户。

  编报时的 README 依据

  _{<em>Tears of Steel</em>, <em>Sintel</em>, and <em>Big Buck Bunny</em> © Blender Foundation, CC BY 3.0.} </p> --- TorrentPlayer is a free, open-source desktop app that **plays any torrent straight from BitTorrent**, video or audio, the way a streaming service would. Press play and it starts in seconds; drag the seek bar to minute 47 and it is playing minute 47 a moment later, long before the download gets there. No separate torrent client, no waiting for the file, no server in between. Use it for whatever you stream from torrents: films and series, music albums, podcasts,

  - **Paste, drop, or search.** Magnet links, `.torrent` files and links (drag one onto the window), or the built-in search. - **Instant start for the best result.** Each search marks a "Best pick" and quietly preloads its first minutes, so pressing play on it starts at once. One preload at a time, never uploaded, deleted if unused, and off with one switch. ### A queue that downloads ahead, intelligently - **Everything you queue downloads in the background** while you watch, so the next video is ready before you get to it. Set a maximum in Settings if you prefer fewer at once.

  | -------------------------------- | ------------------------ | | Windows 10/11 (x64) | `.msi`, `.exe` (NSIS) | | macOS 13+ (Apple Silicon, Intel) | `.dmg`, `.app` | | Linux (x64) | AppImage, `.deb`, `.rpm` | The player uses your system's web engine (WebView2 on Windows, WebKit on macOS and Linux), so it plays the formats that engine supports: commonly MP4, WebM, and MKV with H.264, VP9, or AV1 video and AAC, Opus, or MP3 audio. Dolby (AC-3, E-AC-3, TrueHD) and DTS audio in MP4 and MKV files is converted to AAC as it plays.

  the playback library is a user-only local SQLite database and is never sent anywhere. ## Get TorrentPlayer TorrentPlayer is an early preview. The latest version is **0.3.0**; see the [changelog](CHANGELOG.md) for what changed. You can: - **Download a release** from the [Releases](https://github.com/torrentplayer/torrentplayer/releases) page: installers for Windows, macOS, and Linux with SHA-256 checksums, an SBOM, and build provenance attestations ([how to verify them](docs/release.md#verifying-a-download)). They

  are not code-signed yet, so Windows SmartScreen and macOS Gatekeeper will warn on first launch. - **Download a test build.** Every push to `main` builds unsigned installers for Windows, macOS, and Linux: open the latest [Build workflow run](https://github.com/torrentplayer/torrentplayer/actions/workflows/build.yml) and download the `installers-<platform>` artifact. Unsigned builds trigger SmartScreen on Windows and Gatekeeper on macOS. - **Build from source** (below). | Platform | Packages |

  查看 README 缓存打开 GitHub
• nace-ai/kapitsaC++ · 395 Stars

  模型本地部署、模型接口服务

  Stars 统计更新于 2026/10/24 07:56:07

  查看原始资料

  作者描述

  Blazing fast Apple Silicon optimized inference engine built for Qwen & Nemotron. OpenAI + Anthropic API compatible. Built by folks from nace.ai

  编报时的 AI 中文解读

  Kapitsa 是为 Apple Silicon 设计的本地推理引擎，支持运行 Qwen、Nemotron 和 Flint 模型，并可为编码代理提供 OpenAI 和 Anthropic 兼容的 API。

  编报时的 README 依据

  Kapitsa is an inference engine for Apple silicon that runs Qwen, Nemotron and Recursive's Flint models locally, built for coding agents. It starts opencode, Claude Code, Codex or pi on a local model, and serves OpenAI and Anthropic APIs. [Quickstart](#quickstart) · [CLI](#cli) · [Benchmarks](#benchmarks) · [Models](#models) · [How KFlash works](docs/kflash.md) · [Docs](docs/usage.md) ## 🔥 What's new

  kapitsa serve # chat + OpenAI/Anthropic APIs at http://127.0.0.1:8000 kapitsa claude # Claude Code on Kapitsa kapitsa doctor # this Mac's GPU and memory, and which models fit it kapitsa bench # benchmark this Mac, share it with the community kapitsa bench compare # the same test against the other engines installed here kapitsa pull 27b # download a model without starting it kapitsa list # downloaded models and their size kapitsa rm 27b # delete a downloaded model

  - [2026/10] 📊 **`kapitsa bench compare`: see it on your own Mac.** The same test against every engine installed here, in one command. ## Quickstart **Requirements:** a Mac with Apple silicon (M1 or newer) and macOS 14 Sonoma or later. ```bash brew install nace-ai/tap/kapitsa # or: curl -fsSL https://raw.githubusercontent.com/nace-ai/kapitsa/main/install.sh | bash ``` ```bash kapitsa opencode --model 35b # also: claude, codex, pi ``` _{On a 16-24 GB Mac use `--model 9b`, or `4b` for the quickest replies.} ## CLI ```bash

  | **Nemotron Nano 9B v2** · `nano-9b`<br>_{hybrid Mamba-2 · NVFP4, 7.9 GB} | 🟩 **25**<br>_{no spec} | — | — | — | — | 18 | 14<br>_{Q4_K_M} | — | </details> <sub>Other precisions: `--model 35b:6bit` is five times closer to bf16 than the 4-bit default ([KL](docs/benchmarks.md#quality-qwen36-35b-a3b-against-bf16-2026-10-19)). All but Lightning, Dory, the two Nemotron Nanos, Flint and the 35B's `16gb` read images. Memory: 16 GB for the 4B and 9B models, 32 GB

  for the 27B, 35B, Lightning, Bolt and Flint, 96 GB for Super-VL, 128 GB for Flash-Next. On a 16 or 24 GB Mac, `35b` takes its `16gb` variant: unsloth's 2-bit GGUF, downloaded and imported in one step ([how](docs/usage.md#gguf-files)). [Every model's variants and draft models](docs/usage.md#models-and-downloads)</sub> ## How it works <picture> <source media="(max-width: 600px)" srcset="docs/assets/how-it-works-narrow.svg"> <source media="(prefers-color-scheme: dark)" srcset="docs/assets/how-it-works-dark.svg">

  查看 README 缓存打开 GitHub
• blendi-remade/prompt2rigJavaScript · 363 Stars

  动作生成、3D动画、角色绑定

  Stars 统计更新于 2026/10/24 07:51:02

  查看原始资料

  作者描述

  Type a motion, watch a rigged character perform it: fast video model + GVHMR + live Mixamo retargeting

  编报时的 AI 中文解读

  这是一个将文字描述的动作转换为3D角色动画的工具，通过视频模型生成动作片段，再用GPU提取3D骨骼和手部运动，并实时映射到Mixamo角色上，支持导出到Blender或Unity。适合动画师、游戏开发者或内容创作者快速制作角色动画。

  编报时的 README 依据

  <p align="center"><img src="docs/media/app.png" alt="The prompt2rig app: the generated clip with its pose overlay on the left, Michelle mid side kick on the stage, the take list along the bottom." width="100%"></p> ## Quick start **Needs:** Node 18+ and a [fal](https://fal.ai) API key. For the HQ pass: Windows, an NVIDIA GPU, [`uv`](https://docs.astral.sh/uv/) and `ffmpeg` on PATH. ```bash git clone https://github.com/blendi-remade/prompt2rig && cd prompt2rig cp .env.example .env # add your FAL_KEY

  node server.mjs # → http://localhost:5177 (the live preview works now) ``` **HQ pass (recommended):** ```powershell hq\setup.ps1 # venv, CUDA PyTorch, GVHMR + ~5 GB of checkpoints ``` Download **SMPL-X v1.1 (NPZ)** from [smpl-x.is.tue.mpg.de](https://smpl-x.is.tue.mpg.de/) (free registration) and put `SMPLX_NEUTRAL.npz` in `hq/body_models/smplx/`. **Finger tracking (optional):** WiLoR needs the MANO hand model. Register at [mano.is.tue.mpg.de](https://mano.is.tue.mpg.de), download *Models & Code* (`mano_v1_2.zip`), then: ```powershell

  hq\.venv\Scripts\python hq\setup_hands.py $HOME\Downloads\mano_v1_2.zip # WiLoR weights + MANO ``` Without MANO, fingers fall back to MediaPipe Hands (rougher depth). Then start the worker next to the web server: ```powershell hq\.venv\Scripts\python hq\worker.py # → 127.0.0.1:5178, loads once and stays warm ``` ## Using it - **Generate** from a prompt or a preset. Every clip starts from the same flux-made *performer* (one person, full body, plain backdrop, static camera), which is the input pose estimation likes best.

  celebrations, clear hand shapes. Flips, spins, floor work and crossed or hidden arms confuse any single-camera pose estimator; the presets and the prompt template steer away from them. ## Renders and recordings **Blender.** Turn a saved take into a studio clip (cyclorama, three-point light, slow camera arc), headless: ```powershell & "C:\Program Files\Blender Foundation\Blender 5.2\blender.exe" -b -P tools/render_blender.py -- takes/<take>.glb --side-by-side ``` It writes `renders/<take>.mp4` (1080×1920, or `--res 1920x1080`) and, with `--side-by-side`, the source clip next to the render.

  | Geist Pixel · Instrument Sans · Sometype Mono | SIL OFL 1.1 (bundled in `public/brand/fonts`) | ✅ | | Mixamo characters | Adobe Mixamo terms (loaded from the three.js CDN, not redistributed) | ✅ in your own projects | | FLUX 3 Max, H3 Max Turbo, FLUX.1 schnell | fal and model-provider terms | per those terms | In short: the web app, the live preview and video generation are commercially usable; the **HQ path (GVHMR, SMPL-X, WiLoR, MANO) is research / non-commercial**, so a commercial product would need to replace or license those pieces. Tested on Windows 11 with an RTX 4090.

  查看 README 缓存打开 GitHub
• sit-in/awesome-grokbot-playbooksHTML · 360 Stars

  配方参考、资源导航、模板索引

  Stars 统计更新于 2026/10/24 07:51:02

  查看原始资料

  作者描述

  中文 Grok Bot 玩法与可抄配方清单：作战室、雷达 Routine、额度体检… 由 Grok Bot + Claude 协作维护 | Chinese-first Grok Bot playbooks

  编报时的 AI 中文解读

  这是一个中文优先的 Grok Bot 配方与生态资源清单，收录了 95 个经核实的条目和 10 个可抄配方，涵盖多 Bot 作战室、Routine 额度体检、X 雷达等玩法，适合想快速上手或寻找现成模板的中文用户。

  编报时的 README 依据

  | **95** | 个条目，每条都用 GitHub API 或页面核过：仓库存在、还在更新、确实和 Grok Bot 相关；许可证情况写在短评里 | | **10** | 个可抄配方，统一格式：场景 → 输入 → 步骤 → prompt → 产出 → 坑，样例全部脱敏 | | **3 级** | 验证标记：✅ 真机跑通 · 🔍 读过文档和配置 · 无标记＝只核实过仓库。✅ 只给真跑过的 | | **171** | 篇社区教程和 X 长文精选（[社区教程](docs/community-tutorials.md) 32 + [X 大 V 文章](docs/x-articles.md) 139），加上作者自己的连载索引，每条都核过存活、作者和发布日期 | | **2 个 AI** | 在维护：Grok Bot 每天找料，Claude Code 核实、写短评，人只拍板。协作记录全公开在 [`.collab/`](.collab/README.md) | 和英文生态导航 [awesome-grok-bot](https://github.com/ZeroPointRepo/awesome-grok-bot) 的分工：它覆盖面广；本仓侧重**中文一人公司、多 Bot 作战室、雷达 Routine 和可抄配方**。不拼数量，宁缺毋滥。 ## 🤖 Grok Bot 是什么

  <img src="https://contrib.rocks/image?repo=sit-in/awesome-grokbot-playbooks" alt="Contributors"> </a> ## 📈 Star 趋势 觉得有用就点个 ⭐，新配方和新条目会持续更新。 <a href="https://star-history.com/#sit-in/awesome-grokbot-playbooks&Date"> <img src="https://api.star-history.com/svg?repos=sit-in/awesome-grokbot-playbooks&type=Date" alt="Star History" width="600"> </a> ## 声明 本仓是社区整理，**不是** Grok Bot 客户端，与 xAI、Cursor 没有任何关联，也未获其认可。Grok、Grok Bot 等名称归各自所有者。条目短评是编者观点，用之前请以官方说明为准。 作者 X：[@sitinme](https://x.com/sitinme)。作者的其他项目：[hiapi.ai](https://hiapi.ai)，面向开发者与 Agent 的图片、视频、音频生成 API 聚合网关。 ## License

  查看 README 缓存打开 GitHub
• 7API/uncensored-ai-api未标注语言 · 222 Stars

  模型接口服务、模型切换、用量统计

  查看原始资料

  作者描述

  Uncensored AI API and models — one key for Claude, GPT, Gemini, Grok, DeepSeek, GLM and Kimi, with uncensored and abliterated variants where available. OpenAI- and Anthropic-compatible. Pay per token.

  编报时的 AI 中文解读

  7API 是一个提供多种前沿 AI 模型（如 Claude、GPT、Gemini 等）的 API 网关，通过一个密钥和兼容 OpenAI 与 Anthropic 的接口，让开发者能够调用包括无审查变体在内的模型，按 token 付费。

  查看 README 缓存打开 GitHub
• appariciojunior/video-editor-skills未标注语言 · 169 Stars

  视频剪辑、字幕添加、音乐音效

  查看原始资料

  作者描述

  Skills that teach Claude Code to be a first-pass video editor in DaVinci Resolve: transcript cuts, vlog pacing, B-roll, music & SFX, clean captions, motion graphics, YouTube export. CC BY-NC 4.0 (no commercial use).

  编报时的 AI 中文解读

  这是一套让Claude Code学习视频剪辑的Skills包，能够辅助YouTube vlog和课程视频的初剪，包括转场、节奏调整、B-roll、音乐与字幕等。面向使用DaVinci Resolve Studio的剪辑新手和创作者，通过读取Skill指令生成脚本，不包含代码。

  查看 README 缓存打开 GitHub
• 7API/glm-5.3-abliterated未标注语言 · 155 Stars

  模型接口服务、提示词合集

  查看原始资料

  作者描述

  GLM 5.3 Abliterated or Uncensored AI model — 35 jailbreak prompts, model guide and API. One 7API key also covers Claude, GPT, Gemini, Grok, DeepSeek and Kimi.

  编报时的 AI 中文解读

  提供GLM 5.3 abliterated/uncensored模型的说明、35个jailbreak提示词，以及通过7API访问模型的指南。适合需要减少模型拒绝行为的开发者和研究人员。

  查看 README 缓存打开 GitHub
• Spectrmachina/ZtrustPython · 153 Stars

  依赖漏洞检测、秘密泄露检测、恶意包检测

  查看原始资料

  作者描述

  Zero-trust supply-chain security for open source: detects malicious & vulnerable dependencies, typosquats, leaked secrets and unsafe GitHub Actions — and pays independent researchers in shielded Zcash via a front-running-proof bounty protocol.

  编报时的 AI 中文解读

  ZTrust 是一个零信任软件供应链安全扫描器，用于检测恶意和易受攻击的依赖、拼写错误、泄露的秘密和不安全的 GitHub Actions，并支持通过屏蔽的 Zcash 支付赏金给独立研究人员。

  查看 README 缓存打开 GitHub
• alexturkey/voxcpm2-turkce-ses-tasarimiSwift · 149 Stars

  语音合成、土耳其语朗读

  查看原始资料

  作者描述

  Turkish voice design with VoxCPM2 on a 4 GB GPU: describe a voice, pick a variant, read Turkish text. Patched audio.cpp, local UI, 7 synthetic voices.

  编报时的 AI 中文解读

  这是一个在4GB GPU上运行的工具，用于设计土耳其语合成语音并朗读土耳其语文本。用户可以用英语描述声音，选择变体，然后朗读文本。

  查看 README 缓存打开 GitHub
• JoasASantos/php2shell-exploitPython · 137 Stars

  漏洞利用

  查看原始资料

  作者描述

  PHP2Shell exploit framework — CVE-2026-109618/619 PHP session.upload_progress UAF to RCE

  编报时的 AI 中文解读

  针对PHP session.upload_progress UAF漏洞的远程代码执行利用框架，支持多种执行模式，适用于授权安全测试与研究。

  查看 README 缓存打开 GitHub
• ulymxumfqtdi/dbd-external-espPython · 116 Stars

  游戏作弊

  查看原始资料

  作者描述

  DBD external ESP: survivor aura, killer wallhack, skillcheck aimbot, injector for Dead by Daylight, educational cheat source

  编报时的 AI 中文解读

  为《黎明杀机》提供外部ESP、自瞄和透视等作弊功能的工具，仅供教育用途。适用于了解作弊风险并愿意承担封号后果的玩家。

  查看 README 缓存打开 GitHub
• Dr-xu193/syncnote-communityDart · 108 Stars

  笔记管理、任务管理、文献阅读

  查看原始资料

  作者描述

  MIT-licensed Flutter workspace for notes, tasks, journals, cloud files and research. Bring your own services.

  编报时的 AI 中文解读

  跨平台个人知识管理工作台，融合笔记、任务、日记、文献阅读、AI 对话及模拟股票研究，适合个人用户整合日常记录与深度研究。

  查看 README 缓存打开 GitHub
• rolottr/bambu-barSwift · 106 Stars

  3D打印监控、打印控制、状态通知

  查看原始资料

  作者描述

  Native macOS menu bar app for Bambu Lab 3D printers. Live progress, time left, temperatures, AMS spools and errors, print controls, several printers and desktop widgets. Local network, no account or cloud.

  编报时的 AI 中文解读

  Bambu Bar 是一款原生 macOS 菜单栏应用，用于监控和管理 Bambu Lab 3D 打印机，可显示打印进度、温度、耗材状态，并支持暂停、恢复、停止等控制操作，无需打开 Bambu Studio 或 Bambu Handy。

  查看 README 缓存打开 GitHub
• Tom123223/maimen-xinshiHTML · 93 Stars

  情绪配餐、菜单审核

  查看原始资料

  作者描述

  麦门心事：把一句心事读成一顿麦当劳，再让三个陌生人决定这顿吃不吃。基于麦当劳中国 MCP 的 Skill + 零依赖单文件演示页。

  编报时的 AI 中文解读

  这是一个基于麦当劳中国MCP的Skill和零依赖单文件演示页，将用户心事转化为麦当劳餐单，并由三个虚拟陌生人审核决定是否通过。适用于一个人点单犹豫不决、需要外部约束的人群。

  查看 README 缓存打开 GitHub
• Motphys/Uni-SuperMimicPython · 80 Stars

  运动跟踪

  查看原始资料

  作者描述

  SuperMimic-UniLab: An Independent reproduction of SuperMimic on UniLab for whole-body humanoid motion tracking.

  编报时的 AI 中文解读

  UniSuperMimic 是 SuperMimic 在 UniLab 上的独立复现，用于 Unitree G1 人形机器人的全身运动跟踪，提供训练、评估和视频导出功能，适合研究人形机器人全身控制的研究人员。

  查看 README 缓存打开 GitHub
• josiahg2502-glitch/BrawlRecompiledC++ · 69 Stars

  游戏移植、模组支持

  查看原始资料

  作者描述

  Native PC port of Super Smash Bros. Brawl (static recompilation): 4K, widescreen, gamepads, and Wii SD mods (Project+, Project+Ex, REX, Brawl Minus, Legacy XP). Bring your own disc.

  编报时的 AI 中文解读

  这是一个将《任天堂明星大乱斗X》静态重编译为原生PC移植版的项目，支持4K分辨率、宽屏、手柄和Wii SD卡模组（如Project+、REX等），需要用户自备游戏光盘镜像。

  查看 README 缓存打开 GitHub
• 78ResearchLab/PHP2Shell-Full-Chain-ExploitPython · 69 Stars

  漏洞检测、漏洞验证

  查看原始资料

  编报时的 AI 中文解读

  该仓库提供针对PHP session.upload_progress相关漏洞（CVE-2026-109618/CVE-2026-109619）的检测与缓解指南，包含Snort、Suricata和ModSecurity规则，以及漏洞验证工具exploit.py和Docker环境，适用于安全研究人员在隔离实验室中验证和防御该漏洞。

  查看 README 缓存打开 GitHub
• cookiengineer/actual-php2shellHTML · 66 Stars

  漏洞复现、安全研究

  查看原始资料

  作者描述

  :robot: NO, There is an easy way to exploit this vulnerability! (vibecoded to prove a point)

  编报时的 AI 中文解读

  这是一个针对PHP 8.5.11两个内存破坏漏洞的远程代码执行概念验证（PoC）项目，通过精心构造的HTTP请求触发漏洞，最终执行系统命令。适用于安全研究人员、漏洞分析师和PHP开发者，用于理解漏洞原理和验证修复效果。

  查看 README 缓存打开 GitHub
• cpypwn/mac2jbC · 65 Stars

  USB键盘模拟、自动发送命令

  查看原始资料

  作者描述

  Use your Apple Silicon Mac as a USB HID keyboard to transfer the ns2jb bootstrap and automate post-bootstrap commands on Nintendo Switch 2.

  编报时的 AI 中文解读

  mac2jb 是一个在 Apple Silicon Mac 上通过 USB HID 键盘模拟向 Nintendo Switch 2 的 SmileBASIC 编辑器传输引导代码并自动执行后续命令的工具，适用于已授权测试的设备。

  查看 README 缓存打开 GitHub
• Sam1983Aing/ticket-tearTypeScript · 60 Stars

  3D交互演示

  查看原始资料

  作者描述

  A 3D ticket you can bend, flip and tear off. Verlet paper physics, holographic GLSL foil and a rip sound made in code. Built with Gemini 4 Argon.

  编报时的 AI 中文解读

  这是一个3D音乐会门票交互演示，用户可以用鼠标弯曲、翻转并撕下票根，体验逼真的纸张物理和全息效果。适合对Web 3D交互和创意编程感兴趣的开发者或设计师。

  查看 README 缓存打开 GitHub

查看本期全部收录项目