从字体注音到原生游戏移植：今日开源项目中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。

• google-research/harness2Python · 6,141 Stars

  代理优化、基准测试

  Stars 统计更新于 2026/10/19 07:55:29

  查看原始资料

  编报时的 AI 中文解读

  Harness² 是一个用于递归代理优化的框架，通过任务级和优化级两层递归，在保持模型权重不变的情况下，自动改进代理的系统提示、工具、记忆等组件，适用于需要提升代理在开放世界中处理未见任务能力的场景。

  编报时的 README 依据

  alternatives, and the [Codex executor](docs/SETUP.md#codex-executor). ## Run a benchmark The following example installs LAB and runs one task, then evaluates and reports its result. Model-backed runs incur API costs; start with a small task selection. ```bash harness2-setup lab harness2-check --bench lab harness2 --bench lab --k 1 --mode parallel \ --only-domains antitrust-competition --max-tasks 1 harness2-judge --bench lab --k 1 --mode parallel harness2-report --bench lab --k 1 --mode parallel ``` Sequential recursion runs the same commands with `--mode sequential`. `--k` is

  the task-level recursion budget; see [Experiments](#experiments). Judging and reporting must use the same benchmark, executor, subset, `--k`, `--mode`, and seed as the refinement run. Add `--dry-run` to `harness2` to preview task selection without model calls. | Benchmark | Tasks | Install | Guide | |---|---|---|---| | JobBench | Professional work | `harness2-setup jb` | [JobBench](docs/BENCHMARKS.md#jobbench) | | LAB | Legal work | `harness2-setup lab` | [LAB](docs/BENCHMARKS.md#lab) |

  <img src="assets/components.png" alt="Eight harness components grouped by runtime role: system prompt, guardrails, memory, and skills construct context; tool guidance and scripts define the tool interface; subagents orchestrate delegated work; plugins control middleware." width="100%"> </a> </p> ## Install Use **Linux, Python 3.12, and [uv](https://docs.astral.sh/uv/getting-started/installation/)**. From the repository root: ```bash uv venv --python 3.12 .venv source .venv/bin/activate uv pip install -e '.[render]' harness2-demo --output runs/demo --k 2 ```

  The local demo runs both recursion modes with deterministic editors and real Python executions. It requires no model credentials. For model-backed benchmarks, install **Node.js 20+, the Google Cloud CLI**, and the [benchmark dependencies](docs/SETUP.md#requirements). LAB needs Pandoc and Poppler; WorkBuddy-Bench needs Docker. Then configure your models: ```bash cp .env.example .env # Edit .env with your Vertex AI project and model IDs. gcloud auth application-default login ``` See the [setup guide](docs/SETUP.md) for model settings, authentication

  Executor network access is unrestricted, and Codex runs with its sandbox disabled. Public benchmark rubrics may be accessible online. Before evaluation, configure [network restrictions](docs/EXPERIMENTS.md#web-access) to prevent access to benchmark answers and grading criteria. ## Results and resume Results are saved under `runs/`, organized by benchmark and subset where applicable. They include task summaries, deliverables, adopted harnesses, reflection notes, and evaluation results. Repeat the same command to resume. Use a new `HARNESS2_RUNS_DIR` when changing

  查看 README 缓存打开 GitHub
• Gathin23/tv-vibehackJava · 1,913 Stars

  电视去广告、启动器替换

  Stars 统计更新于 2026/10/19 07:55:30

  查看原始资料

  作者描述

  Your TV, minus the ads. Debloat any Android TV over ADB and give it a clean, tvOS-inspired launcher. Built in an afternoon with Claude Code.

  编报时的 AI 中文解读

  这是一个用于精简Android TV/Google TV系统并安装类tvOS风格启动器的工具，通过ADB禁用预装应用，提供可逆的恢复脚本，适合希望去除电视广告和定制主屏幕的用户。

  编报时的 README 依据

  # tv-vibehack **Your TV, minus the ads.** Debloat any Android TV / Google TV over ADB and give it a clean, tvOS-inspired launcher. Hand the prompt to Claude Code or any coding assistant, answer a few questions about which apps you use, and it does the rest on the real TV. | Stock home screen | After | |---|---| |  |  | Everything is reversible: packages are disabled, never uninstalled, and a restore script puts the stock launcher back. ## Quick start

  Needs JDK 17 and the Android SDK (platform 34, build-tools 34.0.0). On macOS: `brew install openjdk@17 android-commandlinetools && sdkmanager "platforms;android-34" "build-tools;34.0.0"`. ## Talking to the TV by hand ```bash python3 -m venv tools/venv && tools/venv/bin/pip install -r tools/requirements.txt export TV_HOST=192.168.1.50 # your TV's IP tools/venv/bin/python -I tools/tv.py --probe # model, Android version, firmware tools/venv/bin/python -I tools/tv.py 'pm list packages' # one command

  printf 'cmd1\ncmd2\n' | tools/venv/bin/python -I tools/tv.py --batch tools/disable.sh examples/xiaomi-mi-tv-4a/debloat.txt # apply a list tools/restore.sh examples/xiaomi-mi-tv-4a/debloat.txt # undo it, restore stock launcher ``` ### Gotchas we hit - **macOS: `adb connect` says "No route to host"** while the port is clearly open. Recent macOS treats the Google-signed `adb` as its own app under Local Network privacy and silently blocks it. `tools/tv.py` avoids the binary entirely.

  1. On the TV: Settings → Device Preferences → About → press OK on **Build** seven times → back → Developer options → turn on **USB debugging** (and Network/ADB debugging if your TV shows it). Keep the TV on and on the same Wi-Fi as your computer. 2. Open Claude Code (or your assistant of choice) on your computer and paste the prompt from [`PROMPT.md`](PROMPT.md), filling in your TV model. 3. Accept the "Allow USB debugging?" prompt on the TV when told to, then answer the checklist questions (which apps you use, which four go in the dock).

  - **No "wireless debugging" toggle** on some TVs (Xiaomi among them). USB debugging alone opens network ADB on port 5555. - **Chained `pm` commands** (`pm disable a; pm disable b`) may silently stop after the first. Run one per shell call; `--batch` does this. - **Large pushes hang** through the Python client. Use `tools/b64push.py` for APKs. - **Disable the stock launcher only after the new one is installed**, or the Home button lands on a factory test app. - **Third-party launchers** often charge for basics like a custom wallpaper. Hence `launcher/`.

  - **No on-screen keyboard** after switching launchers (seen on Xiaomi): the default input method pointed at a Baidu IME that isn't installed. Fix with `tools/tv.py 'ime set com.google.android.inputmethod.latin/com.android.inputmethod.latin.LatinIME'`. ## Supported TVs Anything running Android TV or Google TV: Xiaomi/Redmi, TCL, Hisense, Sony, OnePlus, Realme, Nvidia Shield and most others. Not Samsung Tizen, LG webOS or Amazon Fire TV (different OS, no ADB launcher replacement). Tested so far on the models in [`examples/`](examples/). ## Contributing

  查看 README 缓存打开 GitHub
• obsidianmd/peleTypeScript · 1,794 Stars

  图表渲染

  Stars 统计更新于 2026/10/19 07:55:30

  查看原始资料

  作者描述

  Renders Mermaid diagrams faster than you'd run from a volcano! 🌋

  编报时的 AI 中文解读

  Pele 是一个将 Mermaid 图表渲染为 SVG 的库，支持 Mermaid 12.1.0 的所有图表类型，适合需要在网页或应用中嵌入图表渲染功能的开发者。

  编报时的 README 依据

  # Pele Pele renders [Mermaid](https://mermaid.js.org) diagrams to SVG. It is small, synchronous, has no runtime dependencies, and uses CSS variables for colors and fonts. Pele supports every diagram type built into Mermaid 12.1.0. Documentation, examples, and a playground are at [pele.run](https://pele.run). ## Install ```sh npm install pele ``` Or load it [from a CDN](https://pele.run/api#using-a-cdn): ```html <script src="https://cdn.jsdelivr.net/npm/pele@1/dist/pele.global.js"></script> <script>

  for (const el of document.querySelectorAll('pre.mermaid')) pele.mount(el, el.textContent); </script> ``` ## Use ```ts import { render } from 'pele'; const { svg } = render(`flowchart LR A[Mermaid text] --> B[SVG]`); document.querySelector('#diagram').innerHTML = svg; ``` In a browser, `mount()` renders into an element, reads fonts from CSS, and adapts the layout to the container width: ```ts import { mount } from 'pele'; mount(document.querySelector('#diagram'), text); ``` See the [API](https://pele.run/api) for options and error handling.

  Pele targets Mermaid 12.1.0 syntax, with its own layout and visual style. Its parsers are tested against Mermaid's parser specs and with differential fuzzing. ## Develop ```sh npm install npm test npm run build npm run bench ``` The documentation site and playground are in `website/` and use Node 24 and pnpm 12. With nvm: ```sh cd website nvm use pnpm install pnpm dev ``` After installing the website dependencies, run `npm run test:website` from the repository root to check the playground. See [Contributing](CONTRIBUTING.md) for parser testing. ## License

  查看 README 缓存打开 GitHub
• youpi-pg/YouPI-PGJavaScript · 1,314 Stars

  支付网关、支付验证

  Stars 统计更新于 2026/10/19 07:55:51

  查看原始资料

  作者描述

  A self-hosted UPI & COD payment gateway with zero per-transaction platform fees, universal JavaScript checkout integration, dynamic QR payments, WhatsApp OTP authentication, and automated payment verification.

  编报时的 AI 中文解读

  YOUPI PG V2 是一个自托管的 UPI 支付网关，面向独立商户，提供零交易费、直接银行结算、WhatsApp OTP 验证、动态二维码支付和自动支付确认功能。 它通过 Android 应用、Gmail 和 Google Messages 观察银行通知，实现支付验证，并支持在线销售和线下柜台两种模式。

  编报时的 README 依据

  - **Online Selling Mode:** UPI Vector QR + Cash on Delivery (COD) with customer pre-purchase verification and decoupled address lifecycle. - **Offline Store Counter Mode:** Designed for physical retail stores. Supports UPI QR, Cash at Till (`CASH_COUNTER`), and Card at POS Terminal (`CARD_COUNTER`) with an interactive real-time cashier till dock and instant SSE customer approval. ### 7. 🎨 100% White-Label Customizer Rebrand the checkout portal completely through the Merchant Admin Panel: - Custom app name, logo URL, and browser favicon.

  ### 2. Configure Environment ```bash cp .env.example .env ``` *(The default `.env` is pre-configured for localhost out of the box).* ### 3. Zero-Config Launch (Auto-Initialized SQLite) On first launch, YOUPI PG automatically provisions `data/youpi_pg.sqlite`, creates the directory if missing, and executes idempotent migrations and baseline seeds. No manual database setup is required: ```bash npm run dev # or npm start ``` That's it! Your entire gateway cluster will spin up: - 🎛️ **Control Center Supervisor:** [`http://localhost:8090`](http://localhost:8090)

  - 📊 **Merchant Admin Panel:** [`http://localhost:6001`](http://localhost:6001) - 🛒 **YOUPI Showcase Demo Store:** [`http://localhost:5000`](http://localhost:5000) - ⚡ **Payment REST API:** [`http://localhost:6100`](http://localhost:6100) - 💳 **Customer Checkout Portal:** [`http://localhost:7000`](http://localhost:7000) ### 4. Run the Full Regression Test Suite Verify all **31 test suites** across packages, apps, and chaos harnesses: ```bash npm test ``` ```text ============================================================ 🧪 YOUPI PG V2 — FULL REGRESSION TEST RUNNER

  - `/order-expired` — 5-minute session elapsed, slot released for queued customers. - `/order-cancelled` — Customer cancelled checkout session. <br/> --- ## ⚡ Quickstart in 60 Seconds ### Prerequisites - **Node.js `>= 20.0.0` LTS** (`node -v`) - Standard Linux or Windows host - *No Docker or external database services required!* ### 1. Clone & Install Dependencies Standard `npm install` works out-of-the-box across all packages and apps without any protocol issues: ```bash git clone https://github.com/gxurav1710/YouPI-PG.git youpi-pg cd youpi-pg npm install ```

  Because standard UPI alerts do not carry arbitrary merchant metadata, multiple concurrent transactions with identical amounts could create attribution confusion. YOUPI PG solves this with a mathematical invariant: - **Only ONE customer can pay on a VPA at a time.** - Other buyers wait in a database-backed FIFO queue with live position streaming. - Guarantees **100% exact payment matching** with zero payment mismatch errors! ### 4. 📱 Native Android Observer APK (`com.youpi.observer`) A lightweight, battery-optimized Android foreground service that captures:

  查看 README 缓存打开 GitHub
• ly05010419/MacNativeGamesC · 702 Stars

  游戏运行

  Stars 统计更新于 2026/10/19 07:26:01

  查看原始资料

  编报时的 AI 中文解读

  在 Apple 芯片 Mac 上原生运行《塞尔达传说 王国之泪》的工具，将用户自有的游戏文件和密钥转换为 macOS 应用，非模拟器。

  编报时的 README 依据

  brew install opus libvpx python@3.14 ``` **2. Download or clone this repository**, then open Terminal in its `MacNativeGames` folder and run the script with your game file and keys: ```sh python3.14 build_app.py ~/Games/totk.xci -k ~/keys/prod.keys ``` For an NSP, supply the shared fonts too: ```sh python3.14 build_app.py ~/Games/totk.nsp -k ~/keys/prod.keys --firmware ~/Games/firmware # Or reuse the six .bfttf files from a previous build: python3.14 build_app.py ~/Games/totk.nsp -k ~/keys/prod.keys --shared-fonts /path/to/TOTK_build/totk/extracted/firmware/shared-fonts ```

  **iPhone and iPad support is coming soon. iPhone 16 or newer models are recommended for the upcoming iPhone version.** ## What you need - A Mac with Apple silicon and **macOS 26** or later (the graphics use Metal 4). 16 GB of memory or more is recommended. - About **25 GB** of free disk space for the work folder, in addition to your game file. Keep the output app on the **same APFS volume** to share the large data files through cloning. If cloning is unavailable or the app is on another volume, allow roughly **17 GB more** for the app copy.

  - Your Tears of the Kingdom **base game 1.0.0** (`0100F2C0115B6000`) as an **XCI or NSP**, and your `prod.keys`. To build **1.1.2 or 1.4.3**, supply its separate cumulative update as NSP or NSZ with `--update`; no intermediate update is required. NSZ is decompressed and its NCA hashes are verified before extraction. DLC and combined packages are unsupported.

  ## Known issues - Supports base 1.0.0, or base plus the separate 1.1.2 or 1.4.3 update NSP/NSZ. Other versions, DLC, combined packages, compressed base games (NSZ/XCZ) and personalized tickets remain unsupported. Mac only. - No online features, no amiibo. - New shaders can cause compilation stutters. Cache reuse reduces these, but does not guarantee stutter-free gameplay.

  查看 README 缓存打开 GitHub
• showheyohtaki/motion-promptJavaScript · 408 Stars

  AI辅助创作、视频剪辑、图像编辑

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  编报时的 AI 中文解读

  Motion Prompt 是一个为 AI 生成内容提供命名和示例的库，包含运动图形、摄像机运动、UI 组件、落地页效果和照片风格等 120 个名称，每个名称都配有动画示例、含义和提示示例。它旨在帮助用户通过准确的名称向 AI 描述所需效果，适用于视频编辑、网页设计和图像处理等场景。

  编报时的 README 依据

  ### 1. Open the library, pick, and send your picks to your AI With the `motion-prompt` skill installed in Claude Code or Codex, asking is enough to open this gallery. You do not need to remember names or numbers. When you are not sure what something is called, have the library opened, the way you would look a word up in a dictionary. ```sh npx skills add showheyohtaki/motion-prompt ``` 1. Ask your AI: ```text Open the motion-prompt library. ```

  Turn the camera move in motion-prompt #032 into an English prompt for Veo. The subject is a coffee cup on a desk. ``` Other ways to install, and what the agent does, are in [Skill reference](#skill-reference) below. ### 2. Look up the name and say it to your AI Open a [list](packs/README.md) or the gallery, find the thing you have in mind, and use its name in your prompt. Adding how long, how strong and what moves makes the result closer to what you want. ```text Bring the title in with a mask reveal from below, 0.6 s, each line slightly after the one before. ``` ```text

  The set from the first post (names of animations for websites and landing pages) is not in this repository yet. The `lp` pack is a different set. Some names overlap with that post (parallax, scroll snap and others), but the selection and the samples were made new for this repository. The `ui` pack is based on "UIパーツ名24" (24 UI part names), posted on X on October 10, and `photo` was also made new for this repository. ## Skill reference Node.js 18 or later is required. ### Install **With a prompt.** Paste this into Claude Code or Codex: ```text

  # Motion Prompt **English** | [日本語](README.ja.md) > 日本語の説明は [README.ja.md](README.ja.md) にあります。 > 中文用户：名称一览目前只有英文和日文。把 [`make-your-own/prompts/en/01-translate.md`](make-your-own/prompts/en/01-translate.md) 交给 Claude Code、Codex 等编程 AI，就可以制作中文或其他语言的版本。 > Other languages: the name lists are in English and Japanese for now. Give [`make-your-own/prompts/en/01-translate.md`](make-your-own/prompts/en/01-translate.md) to a coding AI to make a version in your language.

  查看 README 缓存打开 GitHub
• seedds/nxnativeC · 392 Stars

  游戏运行

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  作者描述

  Switch games running natively on iPhone, iPad and Mac: an NSP folder becomes an unsigned IPA (NSO rewriter, Horizon HLE host, NVN→Metal)

  编报时的 AI 中文解读

  在 iPhone 和 iPad 上原生运行 Nintendo Switch 游戏的工具，将 NSP 文件打包为 IPA 并侧载安装，需要用户自备游戏文件和 prod.keys。

  编报时的 README 依据

  brew install --cask dotnet-sdk python3 -m pip install lz4 capstone fonttools ``` **2. Put one game's NSPs in a folder**, base game and update (and any DLC): ``` ~/Games/MarioWonder/ Super Mario Bros. Wonder [010015100B514000][v0].nsp Super Mario Bros. Wonder [010015100B514800][v327680].nsp ``` **3. Run the script:** ``` python3 scripts/build_ipa_from_nsp.py ~/Games/MarioWonder -k ~/keys/prod.keys ```

  The IPA appears in the same folder (here `~/Games/MarioWonder/SMBWMobile.ipa`). The first build takes 30–60 minutes, later ones about 10. You need free disk space of about twice the NSPs' size plus 10 GB. Options such as `-o` (output path) and `--app-name` are listed by `-h`. **4. Install** the IPA with your sideloading tool (AltStore, Sideloadly, LiveContainer, …). The app gets the game's name, icon and version (the update's version, for example 1.2.1 for Mario Wonder; the base game's when there is no update). ## Play

  | <br>The Legend of Zelda: Link's Awakening | | Needs iOS / iPadOS 26 or later. Other games aren't supported yet. For them the script writes a survey and still tries its best to pack an IPA, guessing the game's settings; such an IPA may not start. ## Pack an IPA **1. Get nxnative and its tools** (once, on a Mac with [Xcode](https://apps.apple.com/app/xcode/id497799835) and [Homebrew](https://brew.sh)): ``` git clone https://github.com/seedds/nxnative.git && cd nxnative brew install xcodegen zstd spirv-cross

  **1. 获取 nxnative 和所需工具**（只需一次，在装有 [Xcode](https://apps.apple.com/app/xcode/id497799835) 和 [Homebrew](https://brew.sh) 的 Mac 上）： ``` git clone https://github.com/seedds/nxnative.git && cd nxnative brew install xcodegen zstd spirv-cross brew install --cask dotnet-sdk python3 -m pip install lz4 capstone fonttools ``` **2. 把一个游戏的 NSP 放进一个文件夹**，本体和更新（有 DLC 也放进去）： ``` ~/Games/MarioWonder/ Super Mario Bros. Wonder [010015100B514000][v0].nsp Super Mario Bros. Wonder [010015100B514800][v327680].nsp ``` **3. 运行脚本：** ```

  - **Short stutters** the first time a new scene appears are normal: the app compiles graphics for it once and remembers them for the next launch. - **Reinstalling:** some sideloading tools give a reinstalled app a new data folder. Copy `save/` and `pso_user.tsv` over from the old one to keep your progress. - **Settings without rebuilding:** put lines like `DM_OPMODE=0` (handheld mode: lower resolution, higher frame rate) in `Documents/env.txt`. ## Known issues - No amiibo and no online features.

  - Fire Emblem Engage: after switching away from the app and back, some sounds may stay silent until you restart it. - Super Mario 3D World: a few small magenta squares can appear near the birds in World 1-1. ## For developers How it works, the script's options, running games on the Mac, adding a game, and per-game notes: [docs/development.md](docs/development.md). In-depth bring-up and debugging notes: [docs/agents](docs/agents/README.md). ## License MIT (see [LICENSE](LICENSE)). Third-party code keeps its own license; see [dm/third_party](dm/third_party/README.md). --- # 中文说明

  查看 README 缓存打开 GitHub
• BeckyGlobal/PaidToStudyAbroadHTML · 344 Stars

  留学奖学金指南、免学费路径、带薪读博信息

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  作者描述

  高性价比人生指南·海外版：拿钱留学。出国留学奖学金、免学费、政府全奖、带薪读博，德国日本学手艺有人发钱，美加澳奖学金单独成节。不是你花钱去留学，是你拿钱去留学。70 条，只引官方来源。关注作者：微信视频号 Becky在地球村

  编报时的 AI 中文解读

  面向中国大陆读者的海外留学奖学金与免学费路径指南，涵盖学手艺、本科硕士、政府全奖和带薪读博等70条建议，每条注明成本、收益、证据等级和官方来源，并附在线检索页。

  编报时的 README 依据

  <div align="center"> <img src="og.png" alt="高性价比人生指南 · 海外版：拿钱留学 —— 不是你花钱去留学，是你拿钱去留学" width="820"> # 高性价比人生指南 · 海外版：拿钱留学 _{An unofficial overseas companion to <a href="https://github.com/eternity4719/HowToLiveBetter">HowToLiveBetter (高性价比人生指南)</a>: scholarships, tuition-free study, fully funded government awards and paid PhDs for readers from mainland China. Official sources only.} 讲怎么不花钱、甚至拿着钱出国学东西。学手艺、学护理，读本科、硕士、博士，蓝领白领都有路。美国、加拿大、澳大利亚的奖学金单独成节。<br> 70 条建议，每条写明你要付出什么、能换回什么、证据有多硬，来源只引各国政府、政府机构、法律原文和公立大学官网。 不用全做：这是一份备选单，不是任务清单。挑一两条够得着的，就算数。

  [](docs/核实记录/) [](#许可) ### [打开在线检索页](https://beckyglobal.github.io/PaidToStudyAbroad/) <table> <tr><td align="right"><b>查阅</b></td><td align="left"> [目录](#目录) · [证据分级](#证据分级) · [核实记录](docs/核实记录/) · [拿钱之前先看绳子](#拿了钱要付出什么几根常见的绳子) </td></tr> <tr><td align="right"><b>检索</b></td><td align="left"> [在线检索页](https://beckyglobal.github.io/PaidToStudyAbroad/)：按国家、按节、按证据等级、按花不花钱筛选全部 70 条

  | 读博能不能当成一份工作，每月拿工资、交社保，毕业还能留下来？ | [4. 读博就是上班：有的国家博士生是员工，按月发工资](book/04-读博就是上班.md) | | 去美国读本科，哪些学校给国际生助学金、录取看不看钱？读研究生，学校和校外基金各有哪些钱？ | [5. 美国：本科和研究生怎么拿奖学金](book/05-美国奖学金.md) | | 去加拿大读本科有哪些国际生大奖和自动奖学金？读研究生，学校资助包、联邦和省政府奖学金怎么拿？ | [6. 加拿大：本科和研究生怎么拿奖学金](book/06-加拿大奖学金.md) | | 去澳大利亚读书，哪些大学给国际生减学费，研究型博士和授课型硕士各有什么奖学金？ | [7. 澳大利亚：本科和研究生怎么拿奖学金](book/07-澳大利亚奖学金.md) | | 哪些名气很大的奖学金不收中国人，哪些国家以前免学费、现在不免了？ | [别白费劲](#别白费劲) | ## 怎么读 - **不用全做**：挑一两条你够得着的就行。语言、学历、年龄，每条的「成本」栏都写了门槛。 - **想按条件挑**：打开[在线检索页](https://beckyglobal.github.io/PaidToStudyAbroad/)，可以按国家、按节、按证据等级筛，也可以按「花不花钱、花多少时间、要不要毅力」筛，几个条件能叠着用。

  奖学金和签证政策每年都改，书里的数字可能过时，也可能有错。真去申请时，以各国政府、使领馆和学校的最新官方说法为准。发现写错了，欢迎[提 issue](https://github.com/BeckyGlobal/PaidToStudyAbroad/issues)。 **关注作者：微信视频号「Becky在地球村」** [](https://beckyglobal.github.io/PaidToStudyAbroad/) [](#目录) [](#证据分级)

  查看 README 缓存打开 GitHub
• deepfounder-ai/desigmaRust · 331 Stars

  界面设计、Figma文件导入、原型制作

  Stars 统计更新于 2026/10/19 07:26:24

  查看原始资料

  作者描述

  Free Figma alternative made on Rust

  编报时的 AI 中文解读

  desigma 是一款用 Rust 编写的免费、开源、离线优先的界面设计工具，支持帧、自动布局、组件、原型、矢量编辑以及真实的 Figma 文件导入，适合需要本地化设计工具的独立设计师和团队。

  编报时的 README 依据

  <h1 align="center"> <img alt="desigma: the design tool you actually own. Free, open source, offline, written in Rust." src="marketing/png/launch-banner-1600x900.png" width="100%"> </h1> <p align="center"> <b>Interface design; an independent, local-first, open-source design tool built in Rust.</b><br> Frames, auto layout, components and variants, prototyping, vector editing and real Figma file import,<br> in a native desktop app written entirely in Rust. Free, offline, and yours to change. </p> <p align="center">

  A deterministic CPU renderer on tiny-skia, a spatial index and a parallel 256&nbsp;px tile cache. Panning and dragging on a 5,000-layer page stay well under a frame. No Electron, no web view. </td> <td width="25%" valign="top"> <h3>🗂️ Figma files in</h3> Import from the Figma REST API JSON or straight from <code>.fig</code> files (the kiwi container, deflate or zstd), with components, instance overrides, styles and prototype links. </td> <td width="25%" valign="top"> <h3>🔤 Open fonts, built in</h3>

  Create a presentation with **File → New deck**. Decks use editable 16:9 frames with slide thumbnails, add/duplicate/delete controls, ordering, overview, and speaker notes. Switch to **Layers** for the existing text, shape, image, and vector editing tools. Notes save with **Save notes** or when navigating to another slide. Slide edits and saved notes support Undo/Redo and persist in the document.

  <img src="docs/images/desigma-dev.png" alt="Dev mode with a balance card selected: a Ready for dev status, an annotation and a dev link, the box model with padding and margin, and SwiftUI code for the card" width="440"> <br> _{Dev mode: status, annotations, measurements and SwiftUI for the selected card.} <h3>Hand off as code</h3> Switch to Dev mode for measurements, a box model and generated code in CSS, Tailwind, SwiftUI, Jetpack Compose or React. Mark frames Ready for dev, and add annotations and links for whoever builds them. <br><br>

  | `desigma-fig` | Figma REST JSON and `.fig` import, REST JSON export | | `desigma-fonts` | Bundled Inter, Fontsource fetch and cache, font substitution | | `desigma-export` | PNG, JPG and SVG export, Dev mode CSS | | `desigma-app` | The egui desktop editor | | `desigma-diff` | Pixel-diff library and CLI | | `desigma-mcp` | Local MCP server for design-file inspection, editing, and code generation | ## Get started ### Download

  ```sh desigma --screenshot out.png --size 1600x1000 --select Card # render one frame and exit desigma --fixture showcase --select Home --size 1600x1000 --scale 2 --screenshot hero.png # the showcase file, rendered at 2x (3200x2000) desigma --select Welcome --rulers --layout-grid columns --zoom 120 --screenshot out.png # rulers + column grid desigma --edit-text "Welcome back" --select-text 0-7 --screenshot out.png # mid-edit with a selection desigma --tab prototype --present # open straight into Present

  screenshot. ## MCP integration Build the local server with `cargo build --locked -p desigma-mcp`, then configure your MCP client to launch `desigma-mcp --workspace /path/to/designs`. It provides 13 tools for importing files, inspecting layers and tokens, generating code, editing with undo/redo, and saving a new `.desigma` file. The server runs an isolated file session; live desktop-editor control is not yet connected. See the [setup and tool reference](crates/desigma-mcp/README.md). ## Native plugin installation

  <img alt="100% Rust" src="https://img.shields.io/badge/100%25-Rust-b7410e?style=flat-square&logo=rust"> <img alt="Windows · Linux" src="https://img.shields.io/badge/Windows%20%C2%B7%20Linux-native-2f7bf5?style=flat-square"> <img alt="License: MIT OR Apache-2.0" src="https://img.shields.io/badge/license-MIT%20%2F%20Apache--2.0-3a3a3a?style=flat-square"> <img alt="Status: early alpha" src="https://img.shields.io/badge/status-early%20alpha-d69e2e?style=flat-square">

  cargo run --release -p desigma-app # the editor, with a demo document cargo run --release -p desigma-app -- file.fig # open a Figma file cargo test --workspace # the test suite ``` On Linux you need the usual windowing headers: `pkg-config libxkbcommon-dev libwayland-dev libx11-dev libxcursor-dev libxrandr-dev libxi-dev libgl1-mesa-dev libfontconfig1-dev`. macOS is built and tested in CI. Want to contribute or cut a release? See [CONTRIBUTING.md](CONTRIBUTING.md) and [docs/releasing.md](docs/releasing.md). Useful flags for automation and testing:

  desigma --fixture variables --panel variables --screenshot v.png # the Variables view desigma-diff compare reference.png actual.png --out diff.png # pixel-compare two images ``` > [!IMPORTANT] > **Status:** desigma is in early alpha. Much of Figma's everyday design surface exists, but **it is not yet a replacement for professional daily work**. The biggest gaps are collaboration, plugins, PDF export, `.fig` features that real files haven't exercised yet, and code-signed release builds. Expect rough edges, and please file issues with your OS, a sample file if you can share one, and a

  - **`.fig` files**, from *File → Save local copy*. desigma unpacks the zip wrapper, decodes the kiwi schema and message, and maps node changes, embedded images, shared styles, instance overrides, layout grids and prototype interactions. The `.fig` format is undocumented. Our decoder follows community reverse engineering and is tested against synthetic files, so expect gaps on real-world files and please report them. ## Under the hood - **One document model:** `desigma-doc` is plain data plus an undo stack. Every edit is a `Command` with an exact inverse.

  查看 README 缓存打开 GitHub
• Ryu7w7/OCTR-NitroFueled-PCC++ · 324 Stars

  游戏运行、在线联机

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  作者描述

  Experimental Online Crash Team Racing: Nitro-Fueled (PS4) running natively on Windows 10/11. Windows fork of SacredFire, built on bbport and shadPS4. No game files included.

  编报时的 AI 中文解读

  这是一个让《古惑狼赛车：氮气加速》PS4版在Windows和Linux上原生运行并支持社区服务器在线游玩的实验性项目。它通过运行时替换PS4系统库和基于shadPS4的渲染器实现，不包含游戏文件，需要用户自行提供游戏及1.21更新的dump。

  编报时的 README 依据

  <p align="center"> <img src="docs/images/logo.png" alt="CTR Nitro-Fueled PC" width="360"> </p> # OCTR: CTR Nitro-Fueled PC, online **Crash Team Racing: Nitro-Fueled (PS4, CUSA13795 v1.21) running natively on Windows 10/11 and Linux, with online play on a community server.** This is the online fork of [GetLostIn's CTR-NitroFueled-PC](https://github.com/GetLostIn/CTR-NitroFueled-PC), the Windows build of [SacredFire](https://github.com/T3CHNOLOG1C/SacredFire), the native Linux port of the game by T3CHNOLOG1C. The game's own x86-64 code runs directly on the CPU, a small

  runtime replaces the PS4 system libraries, and a renderer derived from shadPS4 translates the PS4 GPU work to Vulkan. The Windows layer (loader, memory, TLS, exceptions, files, timers) comes from Supermedo's Windows port of bbport: see [packaging/windows/README.md](packaging/windows/README.md). On top of that, this fork adds **online play** (see [Online](#online) below): public races, private rooms, groups, friends and the Pit Stop through the RyuNET server, with your own nickname, Discord account linking and Discord invites.

  > **No game files are included.** You need your own dump of the game and its 1.21 update. > Both builds of the 1.21 update work: the one for firmware 5.05 (the reference) and the one for > 6.72+ (since v0.2.1). > Not affiliated with or endorsed by Sony Interactive Entertainment, Activision or Beenox. > *Crash Team Racing*, *Nitro-Fueled* and *PlayStation* are trademarks of their owners, named > here only to identify the game this software runs. **Status on Windows: playable.** Menus, Adventure and Arcade, races, sound, cutscenes, Xbox/SDL

  查看 README 缓存打开 GitHub
• antfu/furifontTypeScript · 316 Stars

  日语注音显示

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  作者描述

  A Japanese font that shows furigana by itself.

  编报时的 AI 中文解读

  Furifont 是一款日语字体，通过 OpenType 连字自动在汉字上方显示假名读音，无需 ruby 标记或 JavaScript，适用于支持自定义字体的任何场景。

  编报时的 README 依据

   # Furifont <img src="logo.svg" style="float: right;" alt="Furifont" width="96" height="96"> **Furigana, set by the font.** A Japanese font that shows furigana by itself. The text stays plain: no `<ruby>` markup, no JavaScript, no plugin. It works everywhere that supports custom fonts. Check out the demo and playground on **[furifont.antfu.me](https://furifont.antfu.me)** <!-- eslint-skip --> ```html <p class="furifont">今日は日本語の勉強をします。</p> <style> .furifont { font-family: 'Furifont Sans'; line-height: 1.7; } </style> ``` ## How it works

  ## Download Each [release](https://github.com/antfu/furifont/releases/latest) has Sans and Serif, in Hiragana and Romaji, Regular as OpenType and WOFF2, and a variable font of every weight from ExtraLight to Heavy (`Furifont Sans VF`), as TrueType and WOFF2, whose furigana follows the weight of the text. Set `line-height` to 1.6 or more so the furigana is not clipped. ## Use on websites The [website](https://furifont.antfu.me) has a bookmarklet and a [userscript](https://furifont.antfu.me/furifont.user.js) that set any page in Furifont, with optional okurigana readings. ## Build your own

  Choose another base font, word list, furigana letters or sizes. Needs Node 22.18 or later. ```sh npx furifont dev # Customize: live preview in your browser npx furifont build --base "Source Han Serif JP Bold" --script hepburn --family "My Furigana" ``` Settings live in `furifont.config.json`; their defaults are in [`packages/core/src/defaults.ts`](packages/core/src/defaults.ts). | Package | | | --- | --- | | [`furifont`](packages/cli) | The CLI and the server behind Customize | | [`@furifont/core`](packages/core) | Plans glyphs and OpenType rules from a dictionary |

  Furifont is compiled from Source Han Sans JP or Source Han Serif JP. A build step reads a dictionary of thousands of words and draws the furigana into new glyphs. For each word, the font holds a ligature: 今日 becomes one glyph with きょう above it. Longer words win (日本語 before 日本), and a kanji with no clear reading stays bare. Details on the [website](https://furifont.antfu.me) and in [docs/design.md](docs/design.md). > [!WARNING] > Furigana might be incorrect, due to the limits of fonts: a font sees one run of kanji at a time, not the sentence around it. Keep that in mind when you read.

  查看 README 缓存打开 GitHub
• xiaoyang111222/cn-math-taxonomyJavaScript · 309 Stars

  数学知识图谱、自适应诊断、题库

  Stars 统计更新于 2026/10/19 07:26:04

  查看原始资料

  作者描述

  初中数学知识图谱（课标映射）+ 自适应诊断工具 + 1407 道题库

  编报时的 AI 中文解读

  提供中国初中数学（七至九年级）知识图谱数据、自适应诊断工具和题库，面向教师、学生及教育研究者，用于教学参考、学习诊断和路径推荐。

  编报时的 README 依据

  - 九下 第三十四章（14 个）：随机事件、概率意义、列举法、列表法与树状图、频率估计概率、游戏公平性，以及新增的可能性排序、概率性质、等可能判断、放回与不放回、三步树状图、实际估计、统计图表综合、误解辨析。 - 小节号（如 12.1.2、24.2）是按第一版已有标注推定的，没有逐节核对新教材。新增的 `cnm_305`（箱线图）是否为教材必学内容、`cnm_315`（用随机试验检验游戏的公平性）是补充课题，均已在描述里标明，请教师取舍。 ## 诊断小工具（第四轮；第五轮改为每个知识点最多 2 题） `diagnose/` 是一个基于上述数据的静态、可离线的诊断页面：选一个目标知识点（或整个章节），按先修关系自适应出题（第一题答对则掌握并跳过全部祖先；答错则再出一道同级或更容易的题——第二题也错才算薄弱并回溯到先修，hard 先于 soft；第二题答对标为「不稳」，不回溯），给出薄弱的根因、不稳的知识点、按拓扑序排列的学习路径、相关课标条目原文和一张纯 SVG 路径图；每题作答后显示解析和参考答案；进度保存在 localStorage，可导出 JSON 或打印。核心算法与判题在 `diagnose/engine.js`（纯函数，浏览器与 Node 通用），由 `node tests/test_engine.js` 测试。 ```bash

  # 中国初中数学（七—九年级）知识图谱 · 第一版草稿 用「知识点节点 + 先修边」的有向无环图（DAG）描述人教版初中数学（七上至九下）的知识结构，数据格式参考 [Marble os-taxonomy](https://github.com/withmarbleapp/os-taxonomy)（`topics` + `dependencies`）。 > **状态声明：这是 AI 生成的初稿，未经任何教师审核。** > 知识点的拆分粒度、掌握标准、诊断题答案，尤其是**先修关系（哪些是 hard、哪些是 soft，甚至某条边该不该存在）都需要一线初中数学教师审核**后才能用于教学、诊断或学习路径推荐。诊断题答案虽经抽查心算，但没有逐题由人工复核。 ## 数据来源与对齐依据

  检查：JSON 合法；id 唯一；必填字段齐全；`domain`/`type`/`strength` 取值合法；每条边的两端都存在；无自环、无重复边；无环（有环时打印环路）；每个非外部知识点要么有先修边要么有 `root_reason`；没有完全孤立的节点。并打印按领域、按册次的数量以及边数、hard/soft 数。第三轮新增：`standards` 引用的课标 id 必须存在于 `data/cn-curriculum-standards.json`，`confidence` 只能是 `exact`/`inferred`，课标文件自身的 id 格式与编号连续性；并打印课标条目数、映射覆盖率；正式知识点 `standards` 为空时只打印警告（至多 30 个），设置 `STRICT_STANDARDS=1` 才视为错误。全部通过时输出 `VALIDATION PASSED` 且退出码为 0，否则输出错误并以非 0 退出。 ## 打开浏览页面 页面通过 `fetch` 读取 `data/*.json`，浏览器不允许直接用 `file://` 打开，请在本目录启动一个静态服务器： ```bash python3 -m http.server 8000 # 然后访问 http://localhost:8000/ ```

  题目与答案全部由 AI 生成、未经教师审核，复核者与出题者是同一模型家族；格式、核验方法、统计与局限见 [docs/questions.md](docs/questions.md)。 ## 已知局限 - **审核缺位**：见开头状态声明；先修关系、hard/soft 强度、诊断题难度与区分度都没有经过教师或学生数据检验。 - **覆盖不均**：统计与概率现为 43 个，但全部集中在七下、八下、九下三册（新教材只在这三册安排统计与概率，不含七上、八上、九上）；九下目录为第三方来源推定，非官方、待核（见上）；教材中的“阅读与思考”“数学活动”“小结”“信息技术应用”等栏目没有单独成点，仅少数“综合与实践”“探究与发现”并入。 - **粒度不统一**：个别知识点（如各类应用题）按题型归并，个别（如三角形全等的各判定）拆得较细，需按实际教学再调。 - **题库很小**：每个知识点 4–5 题、诊断时最多问 2 题，题与答案由 AI 生成、未经教师审核，难度与干扰项没有学生数据校准，不能代替完整测评。 - **按教材顺序推断**：部分先修边（如命题、定理类内容）反映的是教材编排而非严格逻辑依赖，已尽量标为 `soft`。 ## 运行校验 ```bash python3 scripts/validate.py ```

  查看 README 缓存打开 GitHub
• assafkip/claudemyPython · 309 Stars

  课程生成、音频转写、网页研究

  查看原始资料

  作者描述

  Claudemy: an open-source alternative to Udemy and Coursera. Give Claude a topic, it researches the web and builds a course of 2-minute audio lessons with cited sources. A free Claude Code skill.

  编报时的 AI 中文解读

  Claudemy 是一个开源的 Claude Code 技能，根据用户指定的主题，通过联网研究生成约12节、每节2分钟的音频课程，并提供课程卡片、练习建议和引用来源，适合希望快速学习特定主题的用户。

  查看 README 缓存打开 GitHub
• alchaincyf/huashu-motionJavaScript · 259 Stars

  视频剪辑、音频编辑、AI辅助编程

  查看原始资料

  作者描述

  让你的agent一键直出商业级产品宣传片：会作曲，能指着画面改。Commercial-grade launch films by your coding agent — point at the frame to change it.

  编报时的 AI 中文解读

  huashu-motion 是一个 Agent Skill，让编码代理（如 Claude Code、Codex）根据产品信息自动生成商业级产品宣传片，包括脚本、配乐、动画，并支持在浏览器编辑器中通过点击画面和评论进行局部修改。

  查看 README 缓存打开 GitHub
• LTSharpe/OblivionRemasteredPerformanceFixC++ · 172 Stars

  游戏性能优化

  查看原始资料

  作者描述

  Engine-level performance fix for Oblivion Remastered (UE5): higher fps in towns, far fewer travel hitches, same image. Source, benchmark and docs.

  编报时的 AI 中文解读

  为《上古卷轴4：湮灭重制版》提供引擎级性能优化，提升城镇和旅行时的帧率，减少卡顿，同时保持画面不变。适用于希望在不牺牲画质的前提下改善游戏流畅度的玩家。

  查看 README 缓存打开 GitHub
• SNO2207/WindowSmashC · 161 Stars

  游戏移植、本地联机

  查看原始资料

  作者描述

  Experimental Windows x64 native porting toolkit for Super Smash Bros. Ultimate: static AArch64 recompilation, a D3D11 runtime, and Local Wireless relay netplay.

  编报时的 AI 中文解读

  这是一个实验性的Windows x64移植工具包，用于将《任天堂明星大乱斗特别版》的AArch64代码静态重编译为Windows可执行文件，并提供D3D11运行时和本地无线联机中继功能。适合希望在自己的Windows PC上运行该游戏并尝试联机的技术用户。

  查看 README 缓存打开 GitHub
• Domainsun/awesome-chinese-secondhand-slang未标注语言 · 144 Stars

  二手暗语查询、避坑指南、合规转译

  查看原始资料

  作者描述

  🐟 中国二手交易暗语全书与避坑合规指南 (闲鱼 / 转转 / 小红书 4,476+ 条实战暗号变体、卖家潜台词透视与买家验货防割指南)

  编报时的 AI 中文解读

  收录闲鱼、转转、小红书等二手交易平台4,569+条实战暗号变体，涵盖114大核心品类，提供卖家潜台词透视、买家避坑验货指南与合规替代词库，适合买家防割韭菜和卖家合规防封号。

  查看 README 缓存打开 GitHub
• Leonxlnx/athenaionJavaScript · 116 Stars

  3D场景浏览、图像渲染

  查看原始资料

  作者描述

  A massive ancient Greek world built entirely in code with three.js. Zero assets, made with Claude Code (Opus 5.5)

  编报时的 AI 中文解读

  这是一个用代码程序化生成的古希腊世界，包含地形、建筑、雕像、植被和水体，可用于浏览和渲染高质量图像与视频。

  查看 README 缓存打开 GitHub
• K0NL/B2recompC · 110 Stars

  游戏重编译

  查看原始资料

  作者描述

  Unofficial static recompilation of Bayonetta 2 (Switch) for Windows. Builds the game on your PC from your own legally dumped copy. No game files, keys or SDK code included.

  编报时的 AI 中文解读

  这是一个将《猎天使魔女2》Switch版静态重编译为Windows原生运行的非官方工具。用户需自行提供合法提取的游戏副本，工具会在本地将游戏代码翻译并编译，并附带重新实现的系统层以支持运行。适合拥有正版游戏并希望在PC上体验的玩家。

  查看 README 缓存打开 GitHub
• kernelpanic0x11/instagram-api-workbenchJava · 101 Stars

  API浏览、API测试

  查看原始资料

  作者描述

  Instagram Android API evidence, OpenAPI catalog, and local JavaScript Swagger workbench

  编报时的 AI 中文解读

  该项目提供本地运行的Swagger工作台，用于浏览和测试从Instagram Android APK静态分析恢复的API接口，适合研究人员和开发者探索Instagram客户端API结构。

  查看 README 缓存打开 GitHub
• mu-zi-lee/TangentryGo · 101 Stars

  内容聚合、阅读辅助、笔记整理

  查看原始资料

  作者描述

  沿着好奇心继续发现 —— 本地优先的多平台内容聚合、阅读与笔记工具

  编报时的 AI 中文解读

  Tangentry 是一款本地优先的多平台内容聚合、阅读与笔记工具，支持将 B站、YouTube、小红书、抖音、小宇宙播客、RSS 和网页文章统一在一个界面中阅读，提供字幕阅读、转录、翻译、笔记和导出功能，适合希望集中管理和深度阅读多平台内容的个人用户。

  查看 README 缓存打开 GitHub
• justxor/deepseek-local-guide-ruShell · 93 Stars

  模型本地部署、模型微调、模型接口服务

  查看原始资料

  作者描述

  🐋 Полный гайд на русском: как установить и запустить DeepSeek локально (V4, R1) на ПК, Mac и сервере — Ollama, llama.cpp, LM Studio, vLLM, Open WebUI, API, RAG, агенты, fine-tuning, лучшие фишки

  编报时的 AI 中文解读

  这是一个面向俄语用户的DeepSeek本地部署完整指南，涵盖从快速入门到高级配置的多个方面，包括模型选择、硬件要求、使用Ollama、llama.cpp等工具，以及API、RAG、微调等进阶内容。适合希望在自己的电脑或服务器上免费、私密地运行DeepSeek模型的用户。

  查看 README 缓存打开 GitHub
• kerrdec97/PS5-Backport-GUI未标注语言 · 78 Stars

  游戏降级、批量处理

  查看原始资料

  作者描述

  PS5 Backport — closed-source early testing beta. Windows x64; Linux and macOS to follow after Windows testing.

  编报时的 AI 中文解读

  PS5 Backport 是一款用于将 PS5 游戏降级到旧版固件的图形界面工具，旨在简化批量处理流程，减少重复手动操作。适用于需要在特定固件版本上运行游戏的玩家和社区测试者。

  查看 README 缓存打开 GitHub
• OAleex/ShaolinMonksRecompC++ · 76 Stars

  游戏移植、高分辨率支持

  查看原始资料

  作者描述

  Native port of Mortal Kombat: Shaolin Monks (Xbox) through static recompilation.

  编报时的 AI 中文解读

  《真人快打：少林武僧》Xbox 版的原生 Windows PC 移植版，通过静态重编译实现，不包含原游戏文件，需自备正版 Xbox 游戏镜像或文件。支持高分辨率、键鼠和可重映射手柄，主线剧情可完整游玩。

  查看 README 缓存打开 GitHub
• QAQLeonard/a7c2-fl23-researchPython · 64 Stars

  相机固件工具、LUT导入、DCP导入

  查看原始资料

  作者描述

  FL2/FL3 porting and custom LUT/DCP import tools for Sony A7C II (ILCE-7CM2) firmware 2.01, with backup, installation, readback verification, and restoration workflows. Includes single-camera test results and known limitations.

  编报时的 AI 中文解读

  为索尼A7C II相机固件2.01提供FL2/FL3移植和自定义LUT/DCP导入工具，包含备份、安装、回读验证和恢复流程，并附有单相机测试结果和已知限制。

  查看 README 缓存打开 GitHub
• yuexuan521/git-to-big-mac-hedgerJavaScript · 60 Stars

  工作量化、套餐推荐

  查看原始资料

  编报时的 AI 中文解读

  这是一个将Git提交记录转化为热量推荐的工具，根据代码增删行数和提交次数评估工作强度，并通过麦当劳MCP接口推荐合适的套餐。适合希望以趣味方式犒赏自己的开发者。

  查看 README 缓存打开 GitHub

查看本期全部收录项目