Browser Fetch

A shared, pluggable real-browser rendering backend (default engine Selenium, headless, optional proxy) that mounts as fetchlib's L3 tier, so multiple skills stop each writing their own Selenium — returns (status, html, headers), normalizes anti-bot challenge pages to status=403 so the waterfall correctly marks them blocked, and exposes register_engine() for opt-in nodriver / camoufox / scrapling. Use when the user needs JS-rendered HTML for a scrape, hits a page that comes back empty or thin without a browser, wants one reusable browser layer across scraping skills, or wants to route a fetch through a proxy. Triggers on "browser-fetch", "要真浏览器渲染", "这站要跑 JS 才有内容", "用 Selenium 抓一下", "挂个浏览器后端", "curl 抓不到内容", "带代理抓", "render JS to scrape", "headless browser fetch", "this site needs a browser". Pairs with /fetchlib (register it as the browser L3 tier, only after curl_cffi / Jina fail) and /api-pacer; deliberately does NOT replace the existing Selenium inside /trustpilot. Honest limit (2026-06, datacenter IP): it t

noique 5227e1a 2 files · 11.3 KB Updated

File contents

noique/cross-border-ecommerce-skills/tree/main/tools/browser-fetch commit 5227e1a049

Frequently asked questions

npx skillmds@latest add noique/browser-fetch