Web Search
Description
Execute a web search and return structured results (title, URL, snippet) for an agent to consume.
When to Use
- Agent needs current information beyond its training cutoff
- Researching a topic or verifying facts
When NOT to Use
- Searching for credentials or private data
- High-frequency automated scraping (respect robots.txt)
Inputs
| Name |
Type |
Required |
Description |
| query |
string |
yes |
Search query |
| num_results |
int |
no |
Number of results (default 5) |
Outputs
| Name |
Type |
Description |
| results |
list |
List of {title, url, snippet} dicts |
Example
import requests
def web_search(query: str, num_results: int = 5, api_key: str = None) -> list:
'''
Uses SerpAPI or DuckDuckGo HTML scraping as fallback.
For production use, get a SerpAPI key.
'''
if api_key:
url = "https://serpapi.com/search"
params = {"q": query, "num": num_results, "api_key": api_key}
data = requests.get(url, params=params).json()
return [
{"title": r.get("title"), "url": r.get("link"), "snippet": r.get("snippet")}
for r in data.get("organic_results", [])[:num_results]
]
# Fallback: return placeholder (avoid scraping in examples)
return [{"title": "Result placeholder", "url": "https://example.com", "snippet": query}]
# results = web_search("Python asyncio tutorial", num_results=3)
# for r in results:
# print(r["title"], r["url"])
ashish993