Reading Websites

Reading Websites

Overview

MANDATORY: Use fetch-url.sh for ALL web content fetching. This REPLACES WebFetch, crawl4ai, and curl.

When to Use

Trigger on ANY of these:

• Domain names: example.com, site.pl, docs.io, github.com/path
• URLs: https://..., http://...
• Keywords: "website", "webpage", "web page", "site content"
• User asks to read/fetch/check/view web content

The Command

~/.claude/skills/reading-websites/fetch-url.sh <url> [char_limit]

Arguments:

• url - Full URL with protocol (e.g., https://example.com/page)
• char_limit - Optional. Max characters to return (default: 512)

Examples:

# Read first 512 chars (default)
~/.claude/skills/reading-websites/fetch-url.sh https://docs.example.com/api

# Read more content
~/.claude/skills/reading-websites/fetch-url.sh https://docs.example.com/api 2000

Why This Wrapper

Security: This script has explicit permission. Raw curl does not.

Do NOT use:

• curl directly
• WebFetch tool
• mcp__crawl4ai tools
• Any other URL fetching method

Why: The wrapper is permissioned and scoped. Other methods require broad permissions that expose security risks.

Common Mistakes

Red Flags - You're About to Violate This Skill

• "curl is simpler for this"
• "WebFetch is already available"
• "crawl4ai returns cleaner output"
• "MCP tools are more capable"
• "I have direct access to web fetching tools"
• "This is just a quick fetch"
• "The wrapper is overkill for this"
• "The skill is a workaround for restricted environments"

All of these mean: Use fetch-url.sh anyway. No exceptions.