Skip to content

Search API Reference

Search and retrieval integrations for augmenting evaluation datasets.

from axion.search import GoogleRetriever, TavilyRetriever, YouRetriever
G

GoogleRetriever

Google Custom Search API integration for web search and retrieval.

T

TavilyRetriever

Tavily AI search API for research-optimized web retrieval with AI summaries.

Y

YouRetriever

You.com search API integration for AI-focused web search results.

GoogleRetriever

axion.search.GoogleRetriever 

GoogleRetriever(api_key: Optional[str] = None, num_web_results: int = 5, crawl_pages: bool = False, max_crawl_tokens: Optional[int] = 10000, **kwargs)

Bases: BaseRetriever

Retriever that uses SerpAPI to perform Google searches and format the results into nodes.

Initialize the GoogleRetriever.

Parameters:

  • api_key (Optional[str], default: None ) –

    SerpAPI key used for authenticating requests. Defaults to the value of the 'SERPAPI_KEY' environment variable if not provided.

  • num_web_results (int, default: 5 ) –

    Number of top search results to return (maximum 20).

  • crawl_pages (bool, default: False ) –

    Whether to fetch and clean full page content from URLs in the search results.

  • max_crawl_tokens (Optional[int], default: 10000 ) –

    Maximum number of tokens to crawl per page (if crawling is enabled).

retrieve async 

retrieve(query: str) -> SearchResults

Perform a search query and return results.

Parameters:

  • query (str) –

    Query to search.

Returns:

  • SearchResults –

    A list of nodes with associated scores.

TavilyRetriever

axion.search.TavilyRetriever 

TavilyRetriever(api_key: Optional[str] = None, endpoint: Literal['search', 'extract', 'crawl'] = 'search', search_depth: Literal['basic', 'advanced'] = 'basic', topic: Optional[str] = 'general', max_results: Optional[int] = 5, crawl_pages: bool = False, max_crawl_tokens: Optional[int] = 10000, days: Optional[int] = None, include_answer: bool = False, include_raw_content: bool = False, include_images: bool = False, include_image_descriptions: bool = False, include_domains: Optional[List[str]] = None, exclude_domains: Optional[List[str]] = None, extract_depth: Literal['basic', 'advanced'] = 'basic', max_depth: Optional[int] = 1, max_breadth: Optional[int] = 20, limit: Optional[int] = 50, instructions: Optional[str] = None, select_paths: Optional[List[str]] = None, select_domains: Optional[List[str]] = None, exclude_paths: Optional[List[str]] = None, allow_external: bool = False, categories: Optional[List[str]] = None, **kwargs)

Bases: BaseRetriever

Retriever for Tavily's Search, Extract, and Crawl APIs.

Initialize the TavilyRetriever with specified parameters.

retrieve async 

retrieve(query: str) -> SearchResults

Retrieve results using the Tavily Search API. Only handles 'search' endpoint. For 'extract' and 'crawl', call their respective methods directly.

extract_url_text 

extract_url_text(url: str) -> str

Extract content from a single URL using the Extract API.

extract async 

extract(url: str) -> SearchResults

Extract content from a URL using the Extract API.

crawl async 

crawl(url: str) -> SearchResults

Crawl content from a URL using the Crawl API.

YouRetriever

axion.search.YouRetriever 

YouRetriever(api_key: Optional[str] = None, endpoint: Literal['search', 'news'] = 'search', num_web_results: Optional[int] = 5, crawl_pages: bool = False, max_crawl_tokens: Optional[int] = 10000, safesearch: Optional[Literal['off', 'moderate', 'strict']] = None, country: Optional[str] = None, search_lang: Optional[str] = None, ui_lang: Optional[str] = None, spellcheck: Optional[bool] = None, **kwargs)

Bases: BaseRetriever

Retriever for You.com's Search and News API.

Initialize the YouRetriever.

Parameters:

  • api_key (Optional[str], default: None ) –

    You.com API key. If not provided, it will attempt to use the YDC_API_KEY environment variable.

  • callback_manager (Optional[CallbackManager]) –

    Optional manager for handling callback events during retrieval.

  • endpoint (Literal['search', 'news'], default: 'search' ) –

    The You.com API endpoint to query — either "search" for web results or "news" for news-specific content. Defaults to "search".

  • num_web_results (Optional[int], default: 5 ) –

    Maximum number of search results to return. Must not exceed 20.

  • crawl_pages (bool, default: False ) –

    Whether to crawl and extract the content of the linked pages from the search results.

  • max_crawl_tokens (Optional[int], default: 10000 ) –

    Maximum number of tokens to retrieve per page when crawling. If None, a default internal value is used.

  • safesearch (Optional[Literal['off', 'moderate', 'strict']], default: None ) –

    Safe search filtering level. Defaults to "moderate" if not specified.

  • country (Optional[str], default: None ) –

    Country code for geo-specific search behavior (e.g., "US" for United States).

  • search_lang (Optional[str], default: None ) –

    Language code to use for the search query (e.g., "en" for English).

  • ui_lang (Optional[str], default: None ) –

    Language code for the UI/localized response (e.g., "en").

  • spellcheck (Optional[bool], default: None ) –

    Whether to enable spell check for the query. Defaults to True if unspecified.

retrieve async 

retrieve(query: str) -> SearchResults

Perform a search query and return results using You.com API.

Parameters:

  • query (str) –

    Query to search.

Returns:

  • SearchResults –

    A list of nodes with associated scores.

Search Integrations Guide Google Search Deep Dive