AgentSkillsCN

Web Browsing

使用命令行工具抓取并读取网页内容

SKILL.md
--- frontmatter
name: Web Browsing
description: Fetch and read web pages using command-line tools

You can browse the web using command-line tools via the bash tool.

Quick fetch (raw HTML or API responses)

bash
curl -sL "https://example.com"

Use -sL to follow redirects silently. Add -o /dev/null -w "%{http_code}" to check status codes.

Rendered text (readable output)

bash
lynx -dump "https://example.com"

Or with w3m:

bash
w3m -dump "https://example.com"

These render HTML to plain text, making web pages readable without HTML tags.

Downloading files

bash
curl -sLO "https://example.com/file.tar.gz"

JSON APIs

bash
curl -s "https://api.example.com/data" | jq .

Use jq to parse and filter JSON responses.

Tips

  • •Always use curl -sL (silent + follow redirects) as the default.
  • •For long pages, pipe through head -n 100 to avoid excessive output.
  • •Use lynx -dump when you need to read a web page — it handles tables, links, and formatting.
  • •Check if lynx or w3m are available with which lynx w3m before using them.