从文档解析到3D场景生成：今日开源项目中的AI工具新动向
日报 · 2026-10-27 · GitHub 新星
覆盖日期：2026-10-27 · 北京时间
发布时间：2026/10/28 08:00:32
本期日报从28个收录项目中精选12个代表性项目，涵盖文档解析、网站集成、AI视频生成、3D场景生成、智能设备改装及创业辅助等多个领域，展现了AI工具生态的多元化发展趋势。
在AI技术快速迭代的当下，开源社区涌现出一批旨在解决实际问题、提升开发效率的工具和项目。本期日报从28个收录项目中挑选了12个具有代表性的项目，它们覆盖了从基础的数据解析、网站交互到创意内容生成、设备个性化改造等多个方面，展现了当前AI工具生态的丰富与活力。
1. 文档解析与数据提取的新高度
文档解析是AI应用的基础环节，也是本期的一个亮点。SoMarkAI/OXR 项目提供了一个开源文档解析模型，能够将PDF和文档图片转换为带有布局坐标的Markdown或JSON结构化数据。该模型宣称在准确率和速度上达到SOTA水平，并在后训练阶段采用了递归自我改进（RSI）循环，这在其领域内尚属首次。OXR最突出的特点是其布局分析模块的独立性与可插拔性，用户可以用自定义模型替换内置的布局模型，同时保留内容识别和阅读顺序功能，这为适应多样化的文档场景提供了极大的灵活性。模型仅有1.3B参数，却能在三个基准测试中排名第一，尤其擅长表格解析，并能识别包括化学方程式、化学结构、代码块在内的13种文档元素类型。
对于有本地部署需求的用户，项目提供了Docker一键启动和源代码运行两种方式，并配有可视化演示界面。值得注意的是，本地运行需要Linux系统、NVIDIA GPU（支持BF16）、CUDA 13.0兼容驱动以及至少24GB显存，推荐48GB以上，这对其硬件环境提出了较高要求。项目提供了免费API和在线Demo供用户体验，但作者明确提示，免费API不保证服务稳定性，请求会排队，不建议用于生产环境。
https://github.com/SoMarkAI/OXR
2. 让AI代理更可靠地操作网站
随着AI代理（Agent）的普及，如何让它们可靠地操作各类网站成为关键需求。Pomerado/pomerado 是一个将网站转换为MCP（Model Context Protocol）集成的工具，用户只需用自然语言描述任务，无需录制HAR文件即可创建集成。其工作原理是“探索网站 → 完成你的示例 → 泛化 → 代理调用”，生成的集成是确定性的，比通用的计算机使用（Computer Use）方式更快更可靠，并且能够处理需要登录和双重认证的网站。
该项目已入选YC F26，采用MIT许可证。它要求macOS或Linux系统、Node 24.21或更高版本，并且需要OpenAI API密钥来构建和审查集成（尽管代理本身可以运行在任何模型上）。对于开发者而言，它能够作为MCP stdio服务器运行，可配置到Claude Code、Codex、Cursor等多种客户端。项目在安全细节上考虑得较为周全：登录信息只在指定网站或认证来源上输入，其他地址会被拒绝；每个mint任务有20分钟的活动时间限制。
https://github.com/Pomerado/pomerado
3. 探索AI驱动的视频与3D内容生成
在创意内容生成领域，两个项目展示了AI的潜力。cth9191/animate 是一个面向Claude Code的过程式动画提示词技能，通过编写代码生成短片动画视频。它支持7种内置风格（如剪纸、素描、Riso印刷、像素等），也允许用户通过提供参考图片、视频或网页来创建自定义风格。其工作流程包含故事检查、外观检查、故事板确认等多个步骤，确保最终产出符合预期。该技能甚至支持使用用户自己的音乐，通过分析节拍进行剪辑，并能让动画与语音配音同步。
另一个项目 deepopen-com/deepopen-3d-webs（光域造景）则是一个图像驱动的3D场景生成系统。用户只需提供一张参考图像和提示词，系统便能沿指定相机轨迹生成可预览、回放和追踪的场景结果，包括色彩/深度动画和3D高斯点云。它内置了多种扩散模型和生成/渲染轨迹选项，适合空间概念设计、风格化内容、设计评审等场景。项目默认以演示模式运行，无需GPU即可浏览和回放官方示例，但要启用完整的3D生成引擎，则需要NVIDIA GPU（建议16GB显存）并下载约30GB的模型权重，缺失任何要求都会自动回退到演示模式。
https://github.com/cth9191/animate
https://github.com/deepopen-com/deepopen-3d-webs
4. 围绕特定平台与场景的辅助工具
针对特定平台和场景的定制化工具同样值得关注。Marker284/yandex-station-extended 是一个关于获取Yandex智能音箱root权限并进行自定义的开放维基，其内容覆盖了获取root、SSH连接、安装软件、连接USB外设等多个方面，甚至包括用自定义LLM替换内置语音助手（如Alice）的教程。该项目为不同型号的Yandex Station设备提供了详细的root方法（如DNS方法、ADB等），并明确标注了哪些型号目前无法破解。项目强调了非官方性质，提醒用户修改固件可能导致保修失效或设备损坏，且仅限用于自己的设备。
对于使用DeepSeek Harness (DSH)的用户，KardDev/dsh-plugins 提供了两个实用插件：一个用于提供英文界面和可编辑动效的UI增强插件，另一个则用于按会话统计API成本，并能按提供商和模型进行细分，还支持可选的定价自动发现功能。这对于需要监控API支出的开发者来说相当实用。项目在Windows上要求使用特定的hoisted依赖布局（nodeLinker: hoisted）以解决esbuild构建问题，这点需要用户留意。
https://github.com/Marker284/yandex-station-extended
https://github.com/KardDev/dsh-plugins
5. 为开发者与创业者提供的“指南针”
除了具体工具，本期还收录了一些指导性的开源项目。blakeandersonw/ai-app-playbook 提供了一套在2026年构建实用且盈利应用的完整指南，从选择问题到实现首个盈利月均有详细说明。作者声称有将5个应用扩展至10万美元月收入的经验，并基于此推荐了一套具体的技术栈，明确列出了“不要使用”的技术（如Vercel、Netlify、Next.js等），并提供了详细的成本估算比较。其建议的流程包括用10个视频和等待名单来验证想法，以及后续的分阶段构建和增长策略。
另一个项目 huaxialucaicaizi/ai-startup-skills 则是一套面向中国一人公司和早期创业者的7步Agent Skill合集，覆盖了从产品验证、编写商业计划书、申请政策、路演、获客到定价、公司注册的完整流程。其核心特点是一个总入口Skill，能自动判断用户所处的创业阶段并路由到对应的子技能，同时实现了步骤间的数据流转（如将验证证据传递给BP撰写）。项目明确表示不提供法律或税务意见，且主要针对中国大陆市场。
https://github.com/blakeandersonw/ai-app-playbook
https://github.com/huaxialucaicaizi/ai-startup-skills
6. 本期观察
综合本期收录的项目，可以看出一个明显的趋势：AI工具正在从通用助手向高度专业化的垂直场景渗透，从文档解析到3D内容生成，从特定硬件改造到创业流程指导，每个项目都在解决一个具体的痛点。同时，这些项目的实用性导向非常明确，许多工具都注重成本控制、部署的简便性以及与其他生态（如MCP、Claude Code）的集成。值得注意的是，项目之间的风格差异也很大，有的追求极致的性能，有的则更注重用户体验与安全保障，这种多样性恰恰反映了开源社区的活力。不过，本期目录中“应用与服务”占多数（17个），而“教程与文档”和“资源合集”相对较少，表明当前社区更倾向于分享立即可用的解决方案。
文章生成于 2026/10/28 08:01:35 · 依据本期保存的资料撰写
本期项目与来源资料
收录数按本期首次达标收录的项目统计。
- SoMarkAI/OXRPython · 1,738 Stars
文档解析
Stars 统计更新于 2026/10/28 07:55:57
查看原始资料
作者描述
An open-source document parsing model with SOTA accuracy and speed, using an RSI loop during post-training.
编报时的 AI 中文解读
OXR 是一个开源文档解析模型，可将 PDF 和文档图片转换为带有布局坐标的 Markdown 或 JSON 数据，适用于需要从文档中提取结构化信息的场景。
编报时的 README 依据
## Quick Start <a id="free-api"></a> ### 0. 🔥 🆓 Free API OXR provides a free API so you can test parsing on your own documents without registering or requesting an API key. We'll keep the free API online until we can no longer afford to run it. ⚠️ Note: We do not recommend using the Free OXR API in production. We do not guarantee service stability, and requests are queued, so processing speed may vary slightly. | Online Demo | https://free-oxr.somark.ai | https://free-oxr.somark.cn | |-|-|-| | API URL | https://free-oxr.somark.ai/api/v1 | https://free-oxr.somark.cn/api/v1 |
modelscope download SoMark/OXR-1.0 --local_dir ./models/OXR-1.0 ``` #### 1.2 Start the Services ##### 1.2.1 Run with Docker (Recommended) We recommend Docker to configure and start all services together. ```Bash docker build -f docker/Dockerfile.allinone -t oxr-1.0 . docker run --rm --gpus all --ipc=host \ -p 127.0.0.1:8000:8000 \ -v "$PWD/models/OXR-1.0:/models/oxr-model:ro" \ oxr-1.0 ``` Once the startup logs show that the services are ready, open http://localhost:8000/demo or http://localhost:8000/docs.
Open http://localhost:8000/demo and upload a PDF or image. You can view the original file, Markdown preview, source, layout bounding boxes, and reading order, and download the Markdown. The demo uses a temporary file cache that expires after 1 hour by default and is renewed during processing. Cached results become inaccessible once the cache expires or is cleared, so download your results promptly. #### 1.4 Send a Parsing Request
somark --base-url https://free-oxr.somark.ai/api/v1 --api-key dummy \ parse document.pdf --formats md,json --out ./parsed-output/ ``` ### 1. Run Locally ```Plain Text Operating system: Linux GPU: NVIDIA (with BF16 support) Driver: NVIDIA driver (compatible with CUDA 13.0) Python: 3.12 (for model downloads / running from source) GPU memory to get started: 24GB on a single GPU (for low-concurrency testing) Recommended GPU memory: ≥48GB on a single GPU (for long documents / higher concurrency) ``` #### 1.1 Download the Code and Models
- Pomerado/pomeradoTypeScript · 1,169 Stars
网站集成
Stars 统计更新于 2026/10/28 07:56:00
查看原始资料
作者描述
YC F26. Turn any website into an API / MCP in minutes. Open source core integration minting harness. MIT.
编报时的 AI 中文解读
Pomerado 是一个将网站转换为 MCP 集成的工具，用户可以用自然语言描述任务，无需录制 HAR 文件即可创建集成，适用于需要让 AI 代理可靠地操作网站的场景。
编报时的 README 依据
Install Pomerado and the Chromium build it drives. ```sh npx -y -p pomerado pomerado-mcp --help npx -y -p pomerado playwright install chromium ``` On Linux, add `--with-deps` after `install` if Chromium's system libraries are missing. Pomerado runs as a standard MCP stdio server. Clients that read an `mcpServers` file use this entry. ```json { "mcpServers": { "pomerado": { "command": "npx", "args": ["-y", "-p", "pomerado", "pomerado-mcp", "mint", "--root", "/absolute/path/to/pomerado-integrations"] } } } ```
| Claude Code | `claude mcp add --scope user pomerado -- npx -y -p pomerado pomerado-mcp mint --root ~/pomerado-integrations` | | Codex | `codex mcp add pomerado -- npx -y -p pomerado pomerado-mcp mint --root ~/pomerado-integrations`, then add `env_vars` | | Gemini CLI | `gemini mcp add -s user -e 'OPENAI_API_KEY=$OPENAI_API_KEY' pomerado npx -y -p pomerado pomerado-mcp mint --root ~/pomerado-integrations` | | Cursor | Add the entry to `~/.cursor/mcp.json` |
│ explore the site → complete your example → generalize → agent calls it │ └──────────────────────────────────────────────────────────────────────────┘ ``` For more details, see [How it works](docs/how-it-works.md). ## Get started Prerequisites - macOS or Linux - Node 24.21 or a later Node 24 release, which you can check with `node --version` - An OpenAI API key, since Pomerado uses OpenAI models to build and review integrations. Your agent can run on any model - An MCP client that runs local servers, such as Claude Code, Codex, Cursor, VS Code, Claude Desktop or Gemini CLI
Pomerado types your login only on the site, or on a sign-in address you pass in `authenticationOrigins` when you call `mint`. A sign-in form whose fields or submit go to any other website is refused before anything is typed. A sign-in whose pages are on another website doesn't publish, even one you pass there. Pomerado doesn't ask about other addresses, and it doesn't check where the page's own scripts send what you typed.
Names start with a lowercase letter and use only lowercase letters, digits and underscores. The integration's folder must not exist yet. Each mint gets 20 minutes of active work, and time spent waiting for your answers doesn't count against it. ## Use your integration When minting finishes, Pomerado saves the integration as a folder of source code and returns its paths. ```text pomerado-integrations/example_reader/ ├── src/ Generated operation modules ├── pomerado.json Entrypoint, input and output schemas, and the questions a run may ask
- DUNHKpcc/Claude-GuardianPython · 1,130 Stars
账号安全、网络配置、环境搭建
Stars 统计更新于 2026/10/28 07:51:10
查看原始资料
作者描述
Claude Guardian is an all-in-one Claude account safety and risk prevention skill.Claude防封实战 It intelligently assesses your device, account history, network environment, and payment options, then provides personalized guidance for local, VPS, or virtual machine setups.
编报时的 AI 中文解读
Claude-Guardian 是一个面向 Claude 用户的账号安全与风险预防 Skill，提供订阅准备、账号注册、网络环境配置、本机/VM/VPS 路线选择等指导，帮助用户降低封号风险。
编报时的 README 依据
- 需要支持 Skills 的 Codex；脚本需要 **Python 3.10+** - Skill 和参考文档可以直接阅读，不运行脚本也能查阅 - 项目不会自行安装 Python、代理工具或后台服务；部分静态 YAML 功能需要可选的 PyYAML，缺少时会停止并说明 macOS / Linux 在 Bash 中运行（已存在仓库时跳过 `git clone`）： ```bash git clone https://github.com/DUNHKpcc/Claude-Guardian.git ( set -eu skill_root="$HOME/.codex/skills" mkdir -p "$skill_root" if [ -e "$skill_root/claude-guardian" ] || [ -L "$skill_root/claude-guardian" ]; then printf '%s\n' '目标已存在，停止安装；请先检查并备份现有 Skill，再手动迁移。' >&2 exit 1 fi cp -R Claude-Guardian/skills/claude-guardian "$skill_root/claude-guardian" ) ```
- **按你的环境调整：**这些教程和脚本作者在自己的 Mac、Windows、VM 和 VPS 上都用过。你的系统和应用版本可能不同，Agent 会先核对、按差异调整，出问题就回滚。 - **没人能保证免封：**价格和准入变化快，购买前看一眼官方页面。 ## 怎么工作 ```mermaid flowchart LR U["你：需求和环境"] --> A["Agent<br/>理解、判断、解释"] S["Skill<br/>经验 · 资料 · 规则"] --> A A --> Q["访谈状态机<br/>问什么 · 附什么资料 · 授权记录"] A --> R["资料库<br/>S01–S08 · R01 · R02"] A --> X["脚本与提示词<br/>由你或经批准后运行"] Q --> A X --> V["验证与回滚"] ``` - **Agent 主导：**理解你的需求，挑资料，解释结果，决定下一步。同一个 Skill 对不同的人给出不同的做法。 - **状态机兜底：**上百个访谈问题按条件调度，选定路线后才问路线相关的问题；每次评估都列出要附上的资料。 - **脚本照文档跑：**VM 搭建、防火墙等脚本由你运行，或经你批准后由 Agent 运行。 ## 分级授权：你不需要每一步都在场
- Jakeschincariol/arena-skillPython · 1,045 Stars
答案优化、AI辅助编程
Stars 统计更新于 2026/10/28 07:35:39
查看原始资料
作者描述
When Claude keeps giving you bad answers, make 100 versions of it fight to the death. Same task, 100 different strategies, a bracket, one answer left. Free Claude Code skill.
编报时的 AI 中文解读
这是一个Claude Code技能，通过让100个AI子代理以不同策略竞争同一任务，选出最优答案。适用于对Claude的回答不满意时，通过淘汰赛机制获得更高质量的解决方案。
编报时的 README 依据
## Use it ``` /arena /arena --quick write the headline for our pricing page /arena --agents 32 fix the flaky test in tests/test_api.py /arena --seed 7 plan my launch week, I have 6 hours a day ``` On its own, `/arena` takes your last request as the task and the answer you did not like as the one to beat. Claude can also reach for it without the slash, when you say something like "that's a bad answer, make them compete". When it starts itself like that, it asks before it spends anything. | flag | what it does | | --- | --- | | `--agents N` | N competitors. Default 100. |
git clone https://github.com/Jakeschincariol/arena-skill.git cp -r arena-skill/skills/arena ~/.claude/skills/ ``` Or as a plugin: ``` /plugin marketplace add Jakeschincariol/arena-skill /plugin install arena-skill@arena-skill ``` Claude Code namespaces plugin skills, so installed as a plugin it shows up as `/arena-skill:arena`. Copy the folder instead if you want plain `/arena`. Project-local instead of global: copy the same folder into your repo's `.claude/skills/`. It needs Claude Code, because it spawns sub-agents with the Agent tool, and Python 3.8 or newer. Nothing to pip install.
against a written rubric that is in this repo for you to read and change. What you get is the strongest answer this tournament found, not a proof that it is right. That is why it shows you the attacks it survived, and why it tells you straight when the answer you rejected scored higher. **100 agents do not all run at the same moment.** Claude Code runs at most 10 tool calls at once by default (the `CLAUDE_CODE_MAX_TOOL_USE_CONCURRENCY` setting), so the arena runs in waves of 10. The spawn is 10 waves, the whole 100-agent run is 70, and it takes a while. If you raise that limit,
winning answer. Applying them is your call, and the skill asks. **The same seed gives the same cards and the same bracket.** Not the same answers. The model is not deterministic. **It does not fix a bad task.** Vague task in, 100 flavours of vague out. ## Files ``` skills/arena/SKILL.md the orchestration steps and the exact brief every sub-agent gets skills/arena/bracket.py the tournament state machine, standard library only skills/arena/strategies.json 15 reasoning modes, 12 workflows, 12 strategies. Edit freely
- cth9191/animateJavaScript · 1,017 Stars
视频生成、动画制作
Stars 统计更新于 2026/10/28 07:35:46
查看原始资料
作者描述
Procedural animation in any style for Claude Code: short single-file canvas videos with story, look and storyboard check-ins. 7 built-in styles or one matched from your references, your own music (beat-mapped), real product screenshots, every format from one piece.
编报时的 AI 中文解读
这是一个用于Claude Code的提示词技能，通过编写代码生成短片动画视频，支持多种内置风格和自定义风格，适用于制作说明、历史故事等。
编报时的 README 依据
Or copy the skill folder by hand: `plugins/animate/skills/animate/` → `~/.claude/skills/animate/`. ## Use ``` /animate a 45-second history of the bicycle, cut paper /animate how a hash map works, in the math style /animate a short about our API in the style of these screenshots: ./refs/ /animate a 20s launch reel for https://my-product.com, cut to ./audio/song.mp3, vertical and square /animate a narrated 45s explainer of how DNS works, math style, 16:9, voice from ElevenLabs ``` Pieces are created in your project under `pieces/<name>/`. You can also drive the tools yourself: ```bash
SKILL=~/.claude/skills/animate # or the plugin's install path node $SKILL/tools/build.mjs pieces/my-piece node $SKILL/tools/tile.mjs pieces/my-piece tile.png 24 48 96 node $SKILL/tools/still.mjs pieces/my-piece 2.5,9,14 frames.png --scale 0.5 node $SKILL/tools/storyboard.mjs pieces/my-piece node $SKILL/tools/beats.mjs pieces/my-piece/audio/track.wav pieces/my-piece --start 12 node $SKILL/tools/capture.mjs https://my-product.com pieces/my-piece/assets/home.png
Requirements: [Claude Code](https://code.claude.com), Node 18+, [Playwright](https://playwright.dev) with Chromium, and ffmpeg on your PATH. Fonts are local system fonts with fallbacks for Windows, macOS and Linux. For voice-overs: Python with `faster-whisper` (word timing) and, optionally, the ElevenLabs connector in Claude (Settings → Connectors) for the voice. ```bash npm i -g playwright && npx playwright install chromium ``` As a plugin: ``` /plugin marketplace add cth9191/animate /plugin install animate@animate ```
- kargulstudio/calender-app-designTypeScript · 606 Stars
日历管理、休假管理
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
编报时的 AI 中文解读
一个基于 Figma 设计稿构建的 Kargul Workspace 人员运营工作区设计原型，主要展示日历和时间休假页面，所有功能均在浏览器内运行，数据存储在本地。
编报时的 README 依据
- **Tailwind CSS v4** with design tokens mapped from the Figma variables - **shadcn/ui** (Radix primitives) themed to the design - **Zustand** for state (`persist` middleware → `localStorage`) - **next-themes** for dark mode, **Sonner** for toasts, **cmdk** for the command palette ## Getting started ```bash npm install npm run dev ``` Open http://localhost:3000. To return to the exact design state, use **Settings → Data → Reset demo data** (or clear site data). ## What works - **Calendar** — week / day / month views; drag on empty space to create, drag
# Kargul Calender Design Prototype A design prototype of **Kargul Workspace**, a people-ops workspace built from the "Dashboard – calendar" Figma design. Live at https://calender-app-design.vercel.app. The **Time off** page (`/time-off`, the default route) reproduces the design pixel-for-pixel; every other page in the sidebar is built in the same visual language. Everything runs in the browser — there is no backend; data persists to `localStorage`. ## Stack - **Next.js 16** (App Router, Turbopack) · **React 19** · **TypeScript**
- Marker284/yandex-station-extendedShell · 499 Stars
智能音箱root、设备改装指南
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
编报时的 AI 中文解读
这是一个关于如何获取Yandex智能音箱root权限并进行自定义的开放维基，内容涵盖获取root、SSH、安装软件、连接USB外设等，面向拥有Yandex Station设备的用户。
编报时的 README 依据
получить root без разборки, закрепить доступ и отключить обновления. Ещё — как поставить свой софт и подключить USB-периферию, и как устроены прошивка, сеть и железо каждой модели. <div align="center"> **⭐ Пригодилось? Поставьте звезду — это мотивирует вести проект дальше.** </div> > [!IMPORTANT] > Пошаговые инструкции живут **на сайте — [hackalice.me](https://hackalice.me)**. Этот README — > только витрина. Всё описанное — для **своих** устройств. ## Что можно сделать - 🔓 **[Получить root](https://hackalice.me/start/#root-by-model)** — у каждой модели свой способ:
DNS-метод, правка CA из factory mode, ADB по USB, приложение для Android. - 🔐 **[SSH и защита от обновлений](https://hackalice.me/start/ssh/)** — чтобы доступ не пропал. - 📦 **[Пакеты](https://hackalice.me/start/packages/)** — `yse` (Entware и `.yap`), [YAM](https://hackalice.me/recipes/yam/) с веб-панелью и Lua-плагинами. - 🔌 **[USB Host](https://hackalice.me/recipes/usb-host/)** — флешки, USB-Ethernet, наушники и ЦАП. - 🎵 **[AirPlay](https://hackalice.me/recipes/airplay/)**, [свои звуки](https://hackalice.me/recipes/sounds/),
<div align="center"> <img src="docs/assets/images/logo.png" alt="Yandex Station Extended" width="140"> # Yandex Station Extended **Открытая вики о том, как получить доступ к умным колонкам Яндекса и что с ними делать дальше** [](https://hackalice.me) [](https://t.me/hackalice)
| Станция Макс | ❌ | [Android-платформа](https://hackalice.me/devices/station-max/) | Актуальная таблица — на сайте: [какой способ для вашей колонки](https://hackalice.me/start/#root-by-model). Нет вашей модели? [Пришлите данные](https://github.com/Marker284/yandex-station-extended/issues/new/choose). ## Что в репозитории ```text docs/ страницы вики (Markdown) — сайт hackalice.me собирается отсюда start/ получение root, SSH, пакеты, восстановление recipes/ рецепты: USB, AirPlay, YAM, LLM… devices/ страницы моделей internals/ прошивка, сеть, сборка софта
Проект неофициальный и не связан с ООО «Яндекс». Модификация прошивки может лишить гарантии и вывести колонку из строя — всё на свой риск и только со своими устройствами. Не публикуйте ключи, токены, MAC-адреса и серийные номера.
- martinkeith700/ruyiJSC++ · 442 Stars
网页自动化、数据采集、JS逆向
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
作者描述
Run any site from Python with the real mini AI headless browser , bypass any security anti-crawler , get the data and token you want
编报时的 AI 中文解读
ruyiJS 是一个面向 AI 驱动自动化的无头浏览器，嵌入真实的 Firefox 引擎（SpiderMonkey 和 Necko），用于从 Python 执行网站 JavaScript、绕过反爬虫验证、进行数据采集和 JS 逆向工程。
编报时的 README 依据
<div align="center"> # 🦊 ruyiJS ### Run any site's JavaScript from Python — with the real Firefox engine. **A new-generation, minimalist browser purpose-built for AI-driven automation.**<br> The authentic Firefox JavaScript engine and networking stack, with no window and no renderer —<br> made for web scraping, JS reverse engineering, data collection, and getting past bot checks.
- **Truly headless.** No window, no layout, no renderer. A pure JavaScript + Web API environment that starts fast and stays light. - **Isolated and reproducible.** Each `Context` is its own process with a persistent JavaScript world and its own proxy, so sessions never leak into each other. - **Offline when you need it.** Serve scripts and responses yourself and replay a whole captured flow with no network. If you scrape, reverse-engineer web JavaScript, or collect data behind a challenge, it's for you. If a plain HTTP client already gets your pages, you don't need it.
Runs on **Windows x64** with **Python 3.10+** (verified on Python 3.14). The wheel bundles the pinned native runtime, so there's nothing else to download. ## 🚀 Quick start **1 · Install.** ```sh python -m pip install ruyijs ``` <sub>The wheel ships the SHA256-pinned SpiderMonkey + Necko runtime; local scripts run offline right after install. No browser download.</sub> **2 · Open a context and run JavaScript.** Variables and functions persist across `eval` calls within the same context. ```python from ruyijs import Context with Context() as context: context.eval("let count = 40")
print(context.eval("count += 2")) # 42 print(context.eval("navigator.userAgent")) ``` **3 · Point it at a real target.** Configure the environment, load the page, and let the site's own JavaScript mint whatever token it demands. ```python with Context(proxy="http://127.0.0.1:7890") as context: context.configure_fingerprint({ "navigator": {"language": "zh-CN", "hardwareConcurrency": 8}, "locale": {"timezone": "Asia/Shanghai"}, }) context.load_page("https://target.example/", html)
ruyiJS provides JavaScript execution and a broad Web API environment on **Windows x64**. It has **no browser window and no full page layout or rendering** — it is a tool for automation, analysis, and research. Use it in compliance with the terms of service of any site you interact with and with applicable law. **Related:** [**ruyiHttp**](https://gitee.com/losenine888/ruyiHttp) — a Requests-style Python HTTP library built on Firefox's native networking components. ## 💬 Contact
- huaxialucaicaizi/ai-startup-skillsPython · 435 Stars
创业流程指导
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
作者描述
创业公司做 AI 产品的 7 步 Agent Skill 合集：验证 → BP → 政策 → 路演 → 获客 → 定价 → 注册。
编报时的 AI 中文解读
面向一人公司和早期创业者的7步Agent Skill合集，涵盖从产品验证到公司注册的完整流程，通过总入口自动路由到对应子技能，并实现步骤间的数据流转。适合在中国做AI工具、应用或Agent的创业者按顺序推进创业流程。
编报时的 README 依据
- [常见问题](#zh-faq) - [贡献与反馈](#zh-contrib) - [作者](#zh-author) - [许可](#zh-license) <a id="zh-intro"></a> ### 一句话介绍 「一人公司工具包」的合集仓库：把 7 个独立的 Agent Skill 打包在一起，再加一个**总入口 skill**。你只要说清楚在做什么、做到哪一步，它会判断你处在创业的哪个阶段，然后调用对应的子 skill。 **痛点** - 一个人做 AI 产品，验证、写 BP、申请政策、路演、获客、定价、注册，每件事都得自己摸索 - 7 个 skill 分开装太麻烦，也不知道先用哪个 - 每一步的产出（验证证据、BP、定价）需要串起来用，而不是各写各的 <a id="zh-steps"></a> ### 7 步工具包 每一步也是一个可以单独安装的仓库： | 步骤 | Skill | 解决什么问题 | |---|---|---| | 1 | [**AI MVP Validation · AI 产品 MVP 验证**](https://github.com/huaxialucaicaizi/ai-mvp-validation) | 先验证有没有人要，再写代码、写 BP |
| 7 | [**OPC Registration & Compliance · 一人公司注册与合规**](https://github.com/huaxialucaicaizi/opc-registration-compliance) | 选主体类型、注册清单、年报与合规日历 | <a id="zh-diff"></a> ### 它和单独安装有什么不同 | 单独安装某一个 skill | 安装 AI 创业 Skill 包 | |---|---| | 需要自己判断该用哪个 | 总入口 skill 按阶段帮你选 | | 每一步的产出各自独立 | 上一步的结果交给下一步用（验证证据 → BP → 政策申请 / 路演） | | 7 个仓库分别安装、分别更新 | 一次安装，`scripts/sync.sh` 一键同步 | 如果你只需要其中一步，直接装对应的单独仓库就行。 <a id="zh-fit"></a> ### 适合 / 不适合 **适合** - 在中国做 AI 工具、AI 应用、Agent 的一人公司和早期创业者 - 想按顺序走完"验证 → BP → 政策 → 路演 → 获客 → 定价 → 注册"的人 **不适合** - 需要正式法律、税务意见：本工具包只提供信息参考，请以官方原文和专业人士意见为准 - 面向 VC 的融资路演：路演教练针对的是园区、孵化器和创业大赛
<a id="zh-quickstart"></a> ### 快速开始 **1. 安装** 克隆本仓库，把 `skills/` 下的 7 个子 skill 和总入口一起放进你的技能目录。 Codex（用户技能目录 `~/.agents/skills`）： ```bash git clone https://github.com/huaxialucaicaizi/ai-startup-skills.git mkdir -p ~/.agents/skills cp -R ai-startup-skills/skills/* ~/.agents/skills/ cp -R ai-startup-skills ~/.agents/skills/ai-startup-skills ``` Claude Code（用户技能目录 `~/.claude/skills`）： ```bash git clone https://github.com/huaxialucaicaizi/ai-startup-skills.git mkdir -p ~/.claude/skills cp -R ai-startup-skills/skills/* ~/.claude/skills/
cp -R ai-startup-skills ~/.claude/skills/ai-startup-skills ``` 其他支持 Agent Skills 的工具（如 Cursor）也可按各自方式安装。规范参考：[OpenAI Codex skills](https://learn.chatgpt.com/docs/build-skills)。 **2. 开始使用** ```text 用 ai-startup-skills 帮我看看我现在在创业的哪一步、下一步该做什么：___（在做什么产品、已经做到哪一步） ``` 也可以直接点名某一步，例如： ```text 用 opc-pricing-sales 帮我给这个 AI 产品定价：___ ``` <a id="zh-usage"></a> ### 怎么用 1. 说清楚你在做什么、已经做到哪一步 2. 总入口判断阶段，最多问两个问题，然后调用对应的子 skill 3. 每一步完成后，它会告诉你下一步建议做什么、为什么 4. 新创业者的默认顺序：1 → 2 → 3 / 7 → 4 → 5 → 6。第 3 步和第 7 步可以同时做，申请政策前通常要先注册主体 <a id="zh-principles"></a> ### 原则 - **不编数据**：不编造行业平均值、案例和用户评价，示例都会标明虚构
│ ├── validate.py # 校验总入口和 7 个子 skill 的 frontmatter │ └── sync.sh # 从 7 个独立仓库重新拉取到 skills/ ├── CHANGELOG.md └── LICENSE ``` <a id="zh-sync"></a> ### 同步与更新 7 个子 skill 的源头是各自的独立仓库。要拉取最新版本： ```bash ./scripts/sync.sh python3 scripts/validate.py ``` <a id="zh-faq"></a> ### 常见问题 **Q：必须 7 步都走一遍吗？**<br> A：不用。总入口会从你还没做的最早一步开始，你也可以直接点名某一步。 **Q：合集和单独仓库的内容一样吗？**<br> A：一样。`skills/` 下的内容由 `scripts/sync.sh` 从独立仓库同步，问题请优先到对应的独立仓库反馈。 **Q：政策和法规信息会过期吗？**<br> A：会。子 skill 只引用官方链接，使用前请点开原文确认最新版本。 **Q：可以用在中国以外的市场吗？**<br> A：验证、BP、路演、定价的思路通用；政策导航、注册与合规、平台获客主要针对中国大陆。
**Q: Can policy and legal info go out of date?**<br> A: Yes. Sub-skills link only to official sources; open the original to confirm the latest version. **Q: Does it work outside China?**<br> A: Validation, BP, pitch and pricing logic is general; policy, registration and platform growth target mainland China. <a id="en-contrib"></a> ### Contributing - Issues with a specific step: open an issue in that standalone repo - Issues with the router or bundle: open an [issue](https://github.com/huaxialucaicaizi/ai-startup-skills/issues) here - Run `python3 scripts/validate.py` before a PR
- KardDev/dsh-pluginsJavaScript · 416 Stars
成本统计、界面增强
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
编报时的 AI 中文解读
DSH 插件合集，包含英文界面、可编辑动效/画布的 UI 插件和按会话统计 API 成本、按提供商/模型拆分的用量统计插件。适用于使用 DeepSeek Harness 并需要增强界面或成本监控的用户。
编报时的 README 依据
# DSH plugins A collection of plugins for **DeepSeek Harness (DSH)**. Each plugin lives in its own directory under `plugins/` and is versioned independently. | Plugin | Description | Status | |---|---|---| | [`dsh-motion-ui-en`](plugins/dsh-motion-ui-en) | English-localized interface, HTML/SVG canvases, editable motion and local MP4 production. Derivative of [dsh-motion-ui](https://github.com/Scorp1o117/dsh-motion-ui). | Active | | [`dsh-cost-usage`](plugins/dsh-cost-usage) | Per-session API cost tracking, provider/model breakdowns and optional pricing auto-discovery. | Active |
## Installing a plugin DSH installs plugins by package spec. Since each plugin lives in a subdirectory here, and package managers resolve Git dependencies at the repository root, install from a checkout: ```sh git clone https://github.com/KardDev/dsh-plugins.git cd dsh-plugins pnpm install dsh plugin add ./plugins/dsh-motion-ui-en # Or install cost tracking into a named profile: dsh plugin --profile <name> add ./plugins/dsh-cost-usage ``` Restart the DSH host and reload the page afterwards. Once a plugin is published to npm, the short form works too: ```sh
dsh plugin add @karddev/dsh-motion-ui-en ``` ## Repository layout ``` plugins/ dsh-motion-ui-en/ # conversational UI, canvases and motion dsh-cost-usage/ # cost tracking and pricing discovery pnpm-workspace.yaml # workspace definition (packages: plugins/*) ``` Plugins are pnpm workspace packages with dependencies hoisted to the repository root (`nodeLinker: hoisted`). The hoisted layout matters on Windows: the plugin build runs esbuild through Node's `child_process`, and a symlinked per-package `node_modules` does not resolve reliably there. ## Working on a plugin ```sh
- blakeandersonw/ai-app-playbookTypeScript · 415 Stars
应用开发指南、技术栈推荐
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
作者描述
How to build useful, profitable apps in 2026. Clone it and point your coding agent at AGENTS.md.
编报时的 AI 中文解读
提供一套2026年构建实用盈利应用的完整指南，涵盖从选型到盈利的各个阶段，并推荐具体技术栈。适合希望快速开发AI应用并实现盈利的开发者。
编报时的 README 依据
| 💪 Get fit and eat better | Cal AI, Ladder, Yuka | | 📵 Take back their time | Opal, Structured | | 📚 Learn | Duolingo, Speak, Photomath | | 🧘 Feel calmer | Hallow, Finch | | 🌱 Understand the world | PictureThis, Yuka | | 💼 Earn | SideShift | ## The path | Phase | What | Check | |---|---|---| | 0 · Niche | Pick a frequent problem AI solves in seconds. Test it with 10 videos and a waitlist | Score ≥ 25/35, real demand | | 1–12 · Build | Brief, repo, backend, auth, AI, web, iOS/Android, payments, analytics, onboarding, retention, launch | "Done when" per phase |
| 13 · First 1,000 | Short-form video, creators on SideShift / Trybe, DMs to influencers and meme pages | Content live and tracked | | 14 · Grow | Weekly: one funnel experiment, one product fix | Weekly report | ## The stack | Job | Use | |---|---| | Backend, data, files, jobs, email | [Cloudflare](https://developers.cloudflare.com/workers/) (Workers, D1, R2, KV, Durable Objects, Queues, Workflows, AI Gateway) | | Auth | [Better Auth](https://www.better-auth.com) | | Database at scale | [PlanetScale](https://planetscale.com) |
| Building | [Claude Code](https://claude.com/claude-code), or [OpenCode](https://opencode.ai) + DeepSeek | **Don't use:** Vercel, Netlify, Lovable, Bolt, v0, Replit hosting, Bubble, FlutterFlow, Firebase, Supabase, Clerk, Auth0, Sentry, PostHog, Next.js, ORMs. ## The money (per month) Benchmark estimate: 30% of downloads start a trial, 30% of trials pay $30/year. That's about $2.70 per download. | | | | | | |---|---|---|---|---| | **Downloads** | 1,000 | 10,000 | 100,000 | 1,000,000 | | **Revenue** | $2,700 | $27,000 | $270,000 | $2,700,000 |
<img src="assets/playbook.png" alt="Useful App Playbook" width="100%"> </p> # How to build useful, profitable apps in 2026 My background: I've scaled 5 apps to $100k+ MRR in the past 3 years. This is the stack I would start with today. We initially built [EcoGPT](https://ecogpt.com) ([apply here](https://ecogpt.com/careers/founding-engineer)) on this stack. Good apps help people get fit, focus, learn, or feel calmer, and people pay for them. This playbook goes from picking a problem to your first profitable month. ## Examples | Help people… | Like | |---|---|
``` Clone https://github.com/blakeandersonw/ai-app-playbook and read AGENTS.md ``` Paste into Claude Code or OpenCode, in an empty folder. Please [follow me](https://github.com/blakeandersonw) and star this repo to stay up to date! [](https://github.com/blakeandersonw) [](https://github.com/blakeandersonw/ai-app-playbook) <p align="center">
- deepopen-com/deepopen-3d-websPython · 381 Stars
三维场景生成
Stars 统计更新于 2026/10/28 07:46:01
查看原始资料
作者描述
Image-driven 3D scene generation system. Users provide a reference image and prompt words, and the system generates previewable, replayable, and traceable scene results according to the camera trajectory; official 3D demonstration assets allow direct viewing of color/depth animations and Gaussian point clouds in the browser 图像驱动的三维场景生成系统
编报时的 AI 中文解读
图像驱动的三维场景生成系统，用户输入参考图像和提示词，生成可预览、回放和追踪的场景结果。适用于空间概念设计、风格化内容、设计评审、教学演示及三维资产创建。
编报时的 README 依据
The system runs in demo mode by default. Without a GPU, the pages, the gallery's 3D demos, and task replay all remain usable. With CUDA and model weights available, you can switch to the full 3D generation engine. ## 2. Use Cases - **Spatial concept design**: Get a look-around version of an interior, building, or natural landscape before deciding whether to invest in refinement - **Stylized content**: Anime, fantasy, holiday, and post-apocalyptic presets load in directly — well suited to style exploration
- **Design reviews**: Put stills, flythrough footage, and point-cloud previews on one page for collaborators - **Teaching and demos**: Demo mode needs no GPU; the official `demo-*` assets work well in classrooms and pitch sessions - **A starting point for 3D assets**: Export PLY Gaussian point clouds for downstream viewing or processing ## 3. Core Features The system has five tabs. **Workbench** Summarizes total tasks, completed and failed counts, sample scene count, run mode, CUDA, disk, host, and service port. Click "Refresh System Data" to re-collect. **Scene Generation**
- Upload a reference image, or load an image, positive/negative prompts, model, seed, steps, and camera trajectory from a Chinese example - Diffusion model options: the standard SD1.5, Blazing Drive V11m, Realistic Vision V5.1, Cartoon Pixar V5 - Generation trajectories: Look Around, Overhead Orbit, Spin in Place 360 - Render trajectories: Push In and Out, Classic Orbit, Subtle Sway - **One-Click Generate** runs preview, color animation, and depth animation end to end - **Step 1 · Generate Gaussians** bakes the scene only and returns a PLY
Generation actions are written to `data/history.json`. Official `demo-*` 3D demo tasks are preloaded. Enter an ID and click "View & Demo" to see task details, the reference image, both animations, the point cloud preview, and the PLY download together. **System Status** Shows host, OS, Python, CUDA, run mode, and the data directory, along with the recommended order of operations.
- Docker Engine 24 or later, with `docker compose` available - Port `8090` open on the local machine or server - 8 GB of RAM or more recommended; official PLY demos need several additional GB of disk - Demo mode (`LD_DEMO_MODE=1`) needs no GPU — the default `Dockerfile` is enough - `LD_DEMO_MODE=0` requires an NVIDIA GPU (16 GB VRAM recommended), the NVIDIA Container Toolkit or a local CUDA environment, `requirements-cuda.txt` / the CUDA build of PyTorch, and compiled `submodules/simple-knn` and `submodules/depth-diff-gaussian-rasterization-min`
- Full mode downloads the diffusion and ZoeDepth weights on startup; allow about 30 GB of cache. If any requirement is missing, it falls back to demo mode automatically ### 6.2 Startup Demo (no GPU / default): ```bash docker compose up -d --build ``` Linux server with GPU: ```bash docker compose -f docker-compose.cuda.yml up -d --build ``` On Windows you can also double-click `start_demo.bat` / `start_cuda.bat`, or follow [RUN\_CUDA.md](RUN_CUDA.md). View logs and stop: ```bash docker compose logs -f docker compose down ``` ### 6.3 Access Local: ```text http://127.0.0.1:8090 ```
- zgx949/ai-pojiaPython · 341 Stars
提示词配置、Skill安装
查看原始资料
作者描述
破甲CodeX。面向小白的实用 Skills 与插件，可直接破甲CodeX发挥出模型最大的潜能，附完整教程和示例。
编报时的 AI 中文解读
为 Codex 用户提供提示词配置和 Skill 的安装工具，附带教程和示例，旨在提升模型在明确授权任务中的执行能力。
- x86rc/unthemidatorC++ · 337 Stars
逆向工程辅助
查看原始资料
作者描述
An automated Themida / Code Virtualizer / WinLicense (v3.0-v4) unpacker and dump fixer tool
编报时的 AI 中文解读
Unpack Themida、Code Virtualizer 和 WinLicense 保护的可执行文件，并修复 PE 结构，使其可运行和调试。 适用于逆向工程师或安全研究人员，需要分析和调试受保护的程序。
- Tencent/WeVisOne-VLJupyter Notebook · 310 Stars
图像分割、目标检测、OCR识别
查看原始资料
作者描述
WeVisOne-VL is a powerful VLM by the WeChat Vision Team at Tencent, unifying object-level and pixel-level perception from points and boxes to masks and alpha mattes.
编报时的 AI 中文解读
WeVisOne-VL 是一个视觉语言基础模型，统一了物体级感知（如检测、指代定位、计数、指向、OCR）和像素级感知（如指代分割、抠图），通过结构化下一词预测和轻量级像素解码器实现从点到框到掩码和alpha抠图的任意粒度定位。适用于需要多任务视觉理解与分割的研究人员和开发者。
- patch-gg/anymovePython · 179 Stars
动作捕捉、动画生成
查看原始资料
作者描述
Motion capture for any creature from one AI video: two cameras in one frame, a DINOv2 joint reader, the rig fitted in both views, a GLB out.
编报时的 AI 中文解读
anymove 是一个将单个 AI 视频转换为任意绑定生物（包括翅膀、尾巴和八条腿）动画的工具，通过双摄像头视图和关节读取器实现 3D 动作捕捉，并导出为 GLB 格式。
- JunWan666/astrbot-for-androidKotlin · 168 Stars
模型本地部署、聊天辅助、模型接口服务
查看原始资料
作者描述
在 Android 手机上本地部署和管理 AstrBot 与 NapCat，内置 PRoot Ubuntu 运行环境，支持原生管理界面、WebUI、日志、备份恢复和独立更新，无需 Root。
编报时的 AI 中文解读
在 Android 手机上本地部署和管理 AstrBot 与 NapCat 的应用，无需 Root，提供原生管理界面、WebUI、日志、备份恢复和更新功能。
- ngxson/esp32-usb-to-airprintC · 157 Stars
无线打印、无线扫描
查看原始资料
作者描述
Add AirPrint / AirScan support to your old printer, using just a cheap ESP32-S3
编报时的 AI 中文解读
将仅支持USB的旧打印机转换为无线打印机和扫描仪，通过ESP32-S3开发板连接打印机并接入WiFi，支持从iPhone、iPad、Mac、Android和Linux设备打印，以及从浏览器或Mac应用扫描。
- Shagg88/COGBoxC++ · 125 Stars
游戏启动器、游戏移植
查看原始资料
作者描述
This is an open source all-in-one launcher for the amazing series on Xbox 360: "Gears of War" franchise. 75% of this amazing series never received a proper remaster, so here I am bringing this project to life! Enjoy!
编报时的 AI 中文解读
这是一个将Xbox 360平台《战争机器》系列游戏移植到Windows PC的非官方启动器，通过静态重编译技术将游戏原生代码转换为Windows程序，支持高分辨率、高帧率、HDR等增强特性。适合拥有正版游戏副本并希望在PC上重温该系列的玩家。
- giantbitmonitor/NextFilee未标注语言 · 118 Stars
文件转换
查看原始资料
作者描述
Filee alternative: NextFilee is a faster, lightweight, and fully offline drag-and-drop file converter.
编报时的 AI 中文解读
NextFilee是一款完全离线的文件转换工具，支持拖拽操作和命令行，可转换图片、文档、PDF、视频、音频和压缩包等多种格式。
- toyxyz/toyxyz_RGBMocapPython · 108 Stars
动作捕捉、角色重定向、FBX导出
查看原始资料
作者描述
Native Windows video motion capture with ExMR, FlowHMR and GVHMR, contact correction, humanoid retargeting, and FBX/VMD export.
编报时的 AI 中文解读
RGB MoCap 是一款原生 Windows 应用，可将视频转换为可编辑的人体动画，支持多人动作捕捉、脚掌/手掌接触修正、角色重定向以及 FBX/VMD 导出，AI 处理在本地运行。
- QuixiAI/QuixiDeskHTML · 99 Stars
聊天辅助、模型服务接入、网页自动化
查看原始资料
编报时的 AI 中文解读
QuixiDesk 是一个用 Rust 编写的原生桌面应用，作为 Claude Desktop 的开源替代品，提供统一的聊天和工作区界面，支持多种 LLM 提供商和内置 MCP 工具。
- Clair-ChenShoujie/Babeldoc4GPT未标注语言 · 90 Stars
PDF翻译、PDF处理
查看原始资料
作者描述
基于 BabelDOC 的 ChatGPT Work 云端 PDF 翻译工具，支持保留版式、双语输出及断点续译，默认无需额外翻译 API Key。
编报时的 AI 中文解读
作为 ChatGPT Work 插件，在云端翻译 PDF，保留公式、表格和版式，支持双语对照和断点续译，适用于阅读外文论文和技术资料的用户。
- derek-zhuolin/motion-skillTypeScript · 88 Stars
视频剪辑、语音转视频
查看原始资料
作者描述
不出镜，用文案或录音把想法做成视频。Script- or voice-driven motion graphics with Codex and Remotion: real footage, captions, editable projects and 96 motion studies.
编报时的 AI 中文解读
Motion Skill 是一套供 Codex 等 Agent 使用的提示词与 Remotion 工程，帮助不想出镜的创作者将文案或录音编排成带动效、字幕的视频。适合不想出镜却能表达观点的创作者、设计师和开发者。
- arantes-gabs/fora-de-cenaPython · 79 Stars
视频通话特效
查看原始资料
作者描述
Saídas criativas para chamadas de vídeo no Windows, com processamento local. Instalador e executável único.
编报时的 AI 中文解读
为Windows视频通话提供创意退出效果的桌面应用，通过手势或手动触发消失效果，所有处理在本地完成，无需付费API。
- typetheoryforall/FVjobsPython · 75 Stars
职位聚合
查看原始资料
作者描述
A Job Board for Formal Verification
编报时的 AI 中文解读
面向形式化验证领域的职位聚合板，集中展示使用Lean、Rocq、Isabelle等证明工具链的岗位，并标注工作授权、薪资等关键信息，适合求职者与招聘方使用。
- PierreTucker147/deepseek-uncensoredPython · 73 Stars
模型接口服务
查看原始资料
作者描述
DeepSeek Uncensored from RouterHub — the website for model guide and API access, with GPT, Claude, GLM, Qwen and Grok under one key.
编报时的 AI 中文解读
提供DeepSeek等模型的API访问指南、去审查模型选择教程及35个免审查提示词文件，面向开发者评估和集成模型。
- JeensLab/AI-CompanionJava · 70 Stars
游戏内AI伙伴、语音交互、资源采集
查看原始资料
作者描述
Олег — ИИ-компаньон для Minecraft 1.21.1 и NeoForge. Общение, голос, добыча, крафт и совместное выживание.