web-search · git:20260416.d3b0b2c · 2026-04-16 · sha256 fe999a63f5a83d8a
web-search git:20260416.d3b0b2cA
Immutable. This exact content never changes and is served at /api/v1/blob/fe999a63f5a83d8a.
No tool to install first
claude plugin marketplace add https://markdownregistry.com/mp/a/art_6nyssiael7f2txlj/fe999a63f5a83d8a2f5ed99e28ba2ff897c30293c815ca45ac4824225fe29ecc.json
Then /plugin install web-search@markdownregistry-6nyssiael7f2txlj in Claude Code. It installs this exact version, the one graded A here, pinned to the commit it was graded at, and leaves it there.
---
name: web-search
description: Search the web and fetch web pages. Use when the user asks to search for information, look something up, or fetch a URL.
metadata:
fastclaw:
always: false
---
# Web Search Skill
Search the web and fetch pages using the available tools.
## Using web_fetch Tool
If the `web_fetch` tool is available, use it to fetch and read web pages:
```
web_fetch(url="https://example.com")
```
## Using exec with Python (fallback)
If web_fetch is not available, use Python with requests:
```python
import subprocess
subprocess.check_call(["pip", "install", "-q", "requests", "beautifulsoup4"])
import requests
from bs4 import BeautifulSoup
# Fetch a page
resp = requests.get("https://example.com", timeout=10)
soup = BeautifulSoup(resp.text, 'html.parser')
# Extract text content
text = soup.get_text(separator='\n', strip=True)
print(text[:2000]) # First 2000 chars
```
## Guidelines
- Prefer the `web_fetch` tool if available — it's faster and cleaner
- When using Python, always install packages first
- Limit output to relevant content — don't dump entire pages
- Summarize long content for the user
- Respect robots.txt and rate limits