菜单
Enterprise-grade web scraping and data extraction skill powered by Crawlee. Features anti-blocking stealth browsers, proxy rotation, and structured JSON output for dynamic web apps.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Benchmark and compare the agentic performance of multiple LLM models on the same task. Use this skill when the user wants to compare models, benchmark LLMs, test which model is better at a task, race models against each other, or evaluate model performance side-by-side. Triggers on phrases like 'benchmark models', 'compare opus vs sonnet', 'which model is better at', 'race these models', 'test model performance', 'agent benchmark'.
A universal social media card and layout generator with automatic AI background removal. Use this when a user wants to remove backgrounds from subjects (images) and composite them into customizable, aesthetic posters for social media (e.g., comparison cards, lookbooks, product showcases). Supports fully customizable titles, subtitles, background images, and text boxes. Now natively supports twin engines satori (default for text-heavy layouts) and pillow (for shadow-heavy 1:1 image comparisons).
全自动矩阵号管家，支持"囤素材模式（生成打包）"与"全自动分发模式（生成过审发帖）"，针对中国社交平台（如小红书）内置防风控、合规洗稿与网感图文渲染流。当用户要求做图文、视频或发布自媒体笔记时触发。
Copywriting detoxification & compliance formatting skill. An essential compliance filter invoked prior to publishing content on domestic social media platforms (such as Redbook, TikTok, WeChat) to defend against image OCR surveillance and banned word detection. Crucial for neutralizing sensitive terms, hardcore tech jargon, and scraping rhetoric.
Generate Redbook (Xiaohongshu) content optimized for the platform's CES algorithm. Use when planning content calendars, writing social media text, or optimizing SEO. Supports diary-style, tutorial, review, and list formats.
Redbook (Xiaohongshu) material rendering and publishing skill. Used to convert markdown text into stunning, natively formatted image cards (cover + content pages) and optionally publish them automatically to the platform. Supports 8 gorgeous CSS themes and 4 smart pagination modes.
| name | crawlee-data-extractor-skills |
| description | Enterprise-grade web scraping and data extraction skill powered by Crawlee. Features anti-blocking stealth browsers, proxy rotation, and structured JSON output for dynamic web apps. |
A powerful, stealthy web crawler built on top of Crawlee. It is designed to handle modern, complex websites (SPA, React/Vue) that aggressively block simple HTTP requests. It uses full browser automation (Playwright/Puppeteer under the hood) disguised as a real user to scrape data cleanly into JSON.
PlaywrightCrawler with stealth plugins to bypass simple bot protections.Since this utilizes the Crawlee Node.js ecosystem, ensure dependencies are installed the first time you run this:
cd scripts/
npm install crawlee playwright
npx playwright install chromium
The core script is scrape.js located in the scripts/ folder. It expects a target URL and an output filename.
node scripts/scrape.js --url "https://news.ycombinator.com/" --output "/tmp/data.json"
For an LLM or Agent, you should invoke this skill whenever the user asks to "scrape", "extract", or "crawl" a specific site, especially if the site relies heavily on JavaScript rendering. After running the tool, read the resulting JSON file and process the data as requested.