majiayu000 Web Search Skill
Role
You are a Web Search Assistant. Use this skill when the user wants to search the web and get content from the results.
When to Use
USE THIS SKILL when the user asks you to:
- Search the web or internet
- Find information online
- Look up documentation, tutorials, or guides
- Research a topic
DO NOT USE for local file search or when the user has the information already.
Installation
The tool is already installed at /home/srose/clawdbot/skills/web-search/
cd /home/srose/clawdbot/skills/web-search
npm install # Already done
Usage
Basic Search
web-search -q "your search query"
With Content Fetching
web-search -q "TypeScript tutorial" --fetch --max-results 3
Options
| Option | Description |
|---|---|
-q, --query |
Search query (required) |
--max-results |
Number of results (default: 5) |
--fetch |
Fetch full page content for each result |
Examples
You: Search for React TypeScript tutorials
Bot: web-search -q "React TypeScript tutorial"
You: Get detailed content from results
Bot: web-search -q "Rust async programming" --fetch --max-results 3
You: Quick answer lookup
Bot: web-search -q "how to install Docker on Raspberry Pi"
Output Format
Search only:
{
"query": "your search",
"results": [
{ "title": "Result Title", "url": "https://...", "snippet": "..." }
]
}
With --fetch:
{
"query": "your search",
"results": [
{
"title": "Result Title",
"url": "https://...",
"snippet": "...",
"content": "Full page content (truncated to ~1500 chars)..."
}
]
}
Notes
- Uses DuckDuckGo HTML (no API key needed)
- Lightweight: uses native
fetch+cheerioparser - No headless browser = very low memory footprint
- Works on Raspberry Pi (arm64) without issues
- ~1-2 seconds per search
- 10 second timeout for fetching page content