Search Tool

Applies to: 4.0.8.1+

Search is powered by ddgs and provides web/news/wikipedia/arXiv retrieval.

1) Initialization

from agently.builtins.tools import Search

search = Search(
    proxy=None,
    timeout=20,
    backend="auto",
    search_backend=None,
    news_backend=None,
    region="us-en",
    options=None,
)

2) Available methods

• search(query, timelimit=None, max_results=10)
• search_news(query, timelimit=None, max_results=10)
• search_wikipedia(query, timelimit=None, max_results=10)
• search_arxiv(query, max_results=10)

3) Use with Agent

from agently import Agently
from agently.builtins.tools import Search

agent = Agently.create_agent()
search = Search(region="us-en")

agent.use_tools([
    search.search,
    search.search_news,
    search.search_wikipedia,
    search.search_arxiv,
])

result = agent.input("Search usage of Agently TriggerFlow").start()
print(result)

4) Output shape

search/search_news/search_wikipedia usually return lists with fields like:

• title
• href or url
• body / snippet

search_arxiv returns:

• feed_title
• updated
• entries[] (title/summary/published/authors/links)

5) Practical tips

• combine with Browse or Playwright for search-then-read flows
• for factual answers, read at least one source page after search snippets