web-search - SKILL.md Agent Skill

name: web-search description: > Search the web using lynx and DuckDuckGo. Performs text-based web searches and can dump specific website content. Only use when the built-in WebSearch tool is not available. triggers: - /web-search - "search the web" - "search for" - "look up online" - "web search" - "find on the internet" - "search duckduckgo" - "search online" - "google this" - "find information about" metadata: author: meain version: "1.0.0" license: MIT

/web-search — Web Search Using Lynx + DuckDuckGo

Search the web using lynx text browser with DuckDuckGo. Bypasses bot protection and provides clean, text-based results.

Note: This skill should only be used when the built-in WebSearch tool is not available and no other more specific search tools exist (e.g., documentation APIs, specialized search tools). If WebSearch is available in the current environment, use that instead.

Usage

# Basic web search
python3 ~/.agents/skills/web-search/scripts/search.py "search query"

# Search with more results
python3 ~/.agents/skills/web-search/scripts/search.py "search query" --limit 10

# Dump specific website content
python3 ~/.agents/skills/web-search/scripts/search.py --dump "https://example.com"

# Search and filter results
python3 ~/.agents/skills/web-search/scripts/search.py "search query" --filter "keyword"

Examples

# Basic search
python3 ~/.agents/skills/web-search/scripts/search.py "best budget earbuds india under 5000"

# Technology news search
python3 ~/.agents/skills/web-search/scripts/search.py "golang 1.24 release notes"

# Get specific website content
python3 ~/.agents/skills/web-search/scripts/search.py --dump "https://www.smartprix.com/mobile_headphones/best-earbuds-under-5000-list"

# Search with filtering
python3 ~/.agents/skills/web-search/scripts/search.py "rust programming tutorial" --filter "beginner"

# Limit results
python3 ~/.agents/skills/web-search/scripts/search.py "machine learning courses" --limit 5

Features

Text-based browsing: Uses lynx to bypass JavaScript and bot protection
Clean output: Extracts readable content without ads and distractions
DuckDuckGo integration: Privacy-focused search engine
Website dumping: Can extract content from specific URLs
Result filtering: Filter results by keywords
Configurable limits: Control number of search results

How It Works

Uses lynx to search DuckDuckGo with your query
Extracts clean search results with titles, descriptions, and URLs
Can follow specific URLs to dump full content
Provides structured, readable output

Requirements

lynx text browser (already available in your environment)
Python 3 (for the search script)

Notes

Lynx bypasses most bot protection mechanisms
Results are text-only (no images or JavaScript content)
DuckDuckGo provides privacy-focused search without tracking
Works well for technical documentation, news, and informational sites
Some heavily JavaScript-dependent sites may not render properly
For fetching individual web pages: Use curl and html2markdown instead of the --dump option for better compatibility and cleaner output