Claude Code 配上这几个 Skills，基本能爬遍全网

• URL: https://x.com/0xecho99/status/2062115773775229318
• Date Saved: 2026-06-05
• Source: X (Twitter Article)
• Tags: dev-tools, ai-engineering

Summary

A curated list of scraping/web-access skills for Claude Code, ordered from light to heavy:

Tools Recommended

1. Agent-reach — 扒社媒（小红书 / X / 抖音 / YouTube / 公众号 / Reddit）

   • Open-source AI scaffold for multi-platform API access
   • github.com/Panniantong/agent-reach

2. Scrapling — 批量抓数据 / 破反爬

   • Adaptive web scraping framework, bypasses Cloudflare, auto-relocates elements after site changes
   • github.com/D4Vinci/Scrapling

3. Browser-use — 扒登录后的页面 / 让AI自己操作

   • LLM-driven browser automation, reuses Chrome login state
   • github.com/browser-use/browser-use

4. Claude in Chrome — 要盯着操作 / 复杂认证页面

   • Anthropic’s official Chrome extension
   • claude.ai/chrome

5. Web-access — 全自动联网 / 接管日常Chrome

   • Agent networking skill, CDP takeover of Chrome with login state, can search bookmarks & history
   • github.com/eze-is/web-access

Principle

从轻到重，够用就停，能不开浏览器就别开。 (Light to heavy, stop when sufficient, don’t open browser if you don’t need to.)

Use Case Mentioned

Author used this stack to scrape a course website — grabbed docs + videos, auto-translated with subtitles, saved into Obsidian.