
以「配置即代码」的方式管理、调试 Clash Verge Rev(mihomo 内核)的配置,并实现不依赖 GUI 的配置更新。涵盖:哪些字段扩展脚本能改 / 不能改(external-controller、secret、端口被内核接管)、改了为什么不生效、如何在纯命令行下让配置重新生成并生效、mihomo external controller 这个内核 RESTful API 的用法、用 external-ui 自托管 Web 面板、以及判断流量「走没走代理 / 命中哪条规则」的排查方法。只要用户在改 Clash Verge / mihomo 的扩展脚本(script)、扩展配置(merge)、订阅规则、external-controller / secret / 端口、或抱怨「clash 配置改了不生效」「规则不命中」「远程连不上面板」「想自动化更新代理配置」,就应该用这个 skill。
Configure and manage Telegram-connected Claude agents with heartbeat scheduling on macOS. Use when the user wants to "set up Telegram agent", "add heartbeat", "configure Telegram channel", "管理 Telegram agent", "配置心跳", "添加定时任务", "Telegram 多 agent", "配置 Telegram", "添加 Telegram agent", "给 Telegram bot 创建身份", "设置定时任务", "新增定时任务", "每天X点执行Y", "schedule a task", "run this automatically", "set up a cron job", or encounters Telegram polling conflicts, 409 Conflict, file upload failures through proxy, or needs to add recurring/periodic automated task execution for Claude agents.
Configure and manage Telegram-connected Claude agents with heartbeat scheduling on macOS. Use when the user wants to "set up Telegram agent", "add heartbeat", "configure Telegram channel", "管理 Telegram agent", "配置心跳", "添加定时任务", "Telegram 多 agent", "配置 Telegram", "添加 Telegram agent", "给 Telegram bot 创建身份", "设置定时任务", "新增定时任务", "每天X点执行Y", "schedule a task", "run this automatically", "set up a cron job", or encounters Telegram polling conflicts, 409 Conflict, file upload failures through proxy, or needs to add recurring/periodic automated task execution for Claude agents.
Shared headed Chrome instance for browser automation. MANDATORY for any scenario requiring a visible (GUI/headed) Chrome browser. This includes but is not limited to: - Social media access: browsing X/Twitter, Reddit, Instagram, posting, reading feeds, verifying tweet dates - News/article verification: checking publish dates, fetching JS-rendered pages, bypassing paywalls - Websites requiring login state: any site where persistent sessions are needed - Anti-bot-protected sites: sites that detect headless/automation browsers - Web form interaction: filling forms, clicking buttons on real websites - Visual page inspection: taking screenshots, checking layouts on live sites NOT required for: headless testing of your own code, PDF generation, Playwright/Puppeteer unit tests. Trigger phrases: "open browser", "check website", "scrape page", "navigate to", "browser automation", "CDP Chrome", "chrome-devtools", "visit URL", "check Twitter/X", "check Reddit", "verify article date", "login to site", "browse", "open page", "access website", "fetch page with JS".
Convert styled HTML files to high-quality single-page PDFs using headless Chrome via playwright-core. Use this skill whenever the user asks to generate a PDF from HTML, export a webpage to PDF, create a no-page-break long-scroll PDF, convert a pitch deck / BP / report HTML to PDF, or mentions "导出 PDF", "生成 PDF", "HTML 转 PDF", "nopagebreak PDF". Also use when iterating on a document with edit-then-export cycles (edit HTML → regenerate PDF → review → repeat). Also use when an interactive/animated HTML presentation needs to be flattened to a static PDF.
将网页文章(含微信公众号等 JS 渲染页面)保存为本地 Markdown 笔记。自动处理图片中的文字信息——对信息图、表格截图等使用 PaddleOCR 提取中英文文本并重构为 Markdown 表格。
将网页导出内容或 PDF 报告重构为可编辑 Markdown。适用于「先改 Markdown 再出 PDF」、修复错误断行、恢复段落结构、补齐图片并按原文位置插入、处理摘要编号与标点风格一致化等任务。
翻译 epub 书籍为中英双语对照版,以及阅读和讨论双语书籍。触发场景:(1) 用户要求翻译一本 epub 为双语对照版("翻译这本书" "做个双语版" "translate this book" "bilingual epub");(2) 用户要求阅读或讨论一本 epub 书籍的内容("读这本书" "讨论这本书" "这本书讲了什么" "summarize this book");(3) 用户提到 epub 文件并想了解内容或添加翻译。
Use when the user asks to "download a paper", "find a paper", "get PDF for DOI", "下载论文", "找论文", "知网下载", or mentions academic paper retrieval needs.
Use when the user asks to "download a paper", "find a paper", "get PDF for DOI", "下载论文", "找论文", "知网下载", or mentions academic paper retrieval needs.
Turn a YouTube link into a polished single-file bilingual (Chinese + original) transcript reading page. Use when the user gives a YouTube URL and asks to "转录" "做转录稿" "生成转录页面" "中英对照" "bilingual transcript" "transcribe this video", or wants a readable HTML transcript with clickable timestamps, chapter navigation, highlighted key points, and proper-noun annotations. Fetches captions + chapters via yt-dlp, the agent translates and curates, then a script renders the HTML.
Use when a user asks the agent to "learn" from a file, example, correction, failed workflow, or feedback and persist that learning into skills or agent instructions. Guides semantic skill refactoring: extract the transferable behavior, update the owning skill so it becomes clearer and easier to execute, avoid append-only note dumping, and decide when not to create new reference files.
Optional per-OS-user headed Chrome provider for browser automation. Use when an environment chooses cdp-chrome for visible GUI Chrome: social media, JS-rendered pages, logged-in sites, anti-bot pages, forms, screenshots, and live site inspection. Not required when an equivalent provider exists.
Shared headed Chrome for browser automation. Mandatory for visible GUI Chrome: social media, JS-rendered pages, logged-in sites, anti-bot pages, forms, screenshots, and live site inspection. Use for requests like open browser, check website, scrape page, visit URL, check X/Twitter or Reddit, verify an article date, or log in to a site. Not for headless tests, PDF generation, or Playwright/Puppeteer unit tests of local code.
Read, navigate, summarize, and reply to WeChat desktop chats on macOS using computer-use MCP. Use for 微信, 群消息, 未读消息, 看群, 读消息, checking unread chats, reading a specific group, viewing shared images, OCR-heavy messages, or pushing WeChat summaries elsewhere.
Configure and manage Telegram-connected Claude agents with heartbeat scheduling on macOS. Use when the user wants to "set up Telegram agent", "add heartbeat", "configure Telegram channel", "管理 Telegram agent", "配置心跳", "添加定时任务", "Telegram 多 agent", "配置 Telegram", "添加 Telegram agent", "给 Telegram bot 创建身份", "设置定时任务", "新增定时任务", "每天X点执行Y", "schedule a task", "run this automatically", "set up a cron job", or encounters Telegram polling conflicts, 409 Conflict, file upload failures through proxy, or needs to add recurring/periodic automated task execution for Claude agents.
将网页文章(含微信公众号等 JS 渲染页面)保存为本地 Markdown 笔记。自动处理图片中的文字信息——对信息图、表格截图等使用 PaddleOCR 提取中英文文本并重构为 Markdown 表格。
Convert styled HTML files to high-quality single-page PDFs using headless Chrome via playwright-core. Use this skill whenever the user asks to generate a PDF from HTML, export a webpage to PDF, create a no-page-break long-scroll PDF, convert a pitch deck / BP / report HTML to PDF, or mentions "导出 PDF", "生成 PDF", "HTML 转 PDF", "nopagebreak PDF". Also use when iterating on a document with edit-then-export cycles (edit HTML → regenerate PDF → review → repeat). Also use when an interactive/animated HTML presentation needs to be flattened to a static PDF.
--- name: wechat-desktop description: Read, navigate, and summarize WeChat (微信) group chat messages on macOS desktop using computer-use MCP. Trigger this skill whenever the user wants to check WeChat messages, read a specific WeChat group, summarize unread WeChat chats, reply in a WeChat group, view images shared in WeChat, or push WeChat summaries elsewhere. Also trigger when the user mentions "微信", "群消息", "未读消息", "看群", "读消息", or any WeChat desktop interaction — even short requests like "帮我看一下微
将网页导出内容或 PDF 报告重构为可编辑 Markdown。适用于「先改 Markdown 再出 PDF」、修复错误断行、恢复段落结构、补齐图片并按原文位置插入、处理摘要编号与标点风格一致化等任务。
翻译 epub 书籍为中英双语对照版,以及阅读和讨论双语书籍。触发场景:(1) 用户要求翻译一本 epub 为双语对照版("翻译这本书" "做个双语版" "translate this book" "bilingual epub");(2) 用户要求阅读或讨论一本 epub 书籍的内容("读这本书" "讨论这本书" "这本书讲了什么" "summarize this book");(3) 用户提到 epub 文件并想了解内容或添加翻译。