Amazon Bedrock AgentCore

Searches the web and returns a ranked list of results. Each result includes a text extract from the page.

You can filter the results by domain and by publication date.

To use this operation, you must have the bedrock-agentcore:InvokeWebSearch permission on the resource arn:${Partition}:bedrock-agentcore:${Region}:aws:tool/web-search.v1.

See also: AWS API Documentation

Request Syntax

client.invoke_web_search(
    webSearchId='string',
    searchQuery='string',
    maxResults=123,
    filters={
        'domainFilter': {
            'include': [
                'string',
            ],
            'exclude': [
                'string',
            ]
        },
        'publishedDateFilter': {
            'publishedAfter': datetime(2015, 1, 1),
            'publishedBefore': datetime(2015, 1, 1)
        }
    }
)

type webSearchId:

string

param webSearchId:

[REQUIRED]

The identifier of the web search tool to invoke. The Amazon Web Services-managed web search tool is web-search.v1.

type searchQuery:

string

param searchQuery:

[REQUIRED]

The text to search for.

type maxResults:

integer

param maxResults:

The maximum number of results to return. Defaults to 25.

type filters:

dict

param filters:

The filters to apply to the search results.

• domainFilter (dict) --

  The domains to include in or exclude from the results.

  • include (list) --

    The domains to return results from. Results from other domains are not returned.

    • (string) --

      A domain name, for example example.com.

  • exclude (list) --

    The domains to never return results from.

    • (string) --

      A domain name, for example example.com.

• publishedDateFilter (dict) --

  The publication time range of the results.

  • publishedAfter (datetime) --

    The earliest publication time of the results, inclusive.

  • publishedBefore (datetime) --

    The latest publication time of the results, inclusive. A date without a time, such as one passed to the AWS CLI, means midnight UTC. To include a whole day, use the last second of that day, for example 2026-01-31T23:59:59Z.

rtype:

dict

returns:

Response Syntax

{
    'results': [
        {
            'title': 'string',
            'url': 'string',
            'text': 'string',
            'publishedAt': datetime(2015, 1, 1)
        },
    ]
}

Response Structure

• (dict) --

  Response for the InvokeWebSearch operation.

  • results (list) --

    The search results, in ranked order.

    • (dict) --

      A web search result.

      • title (string) --

        The title of the page.

      • url (string) --

        The URL of the page.

      • text (string) --

        A text extract from the page.

      • publishedAt (datetime) --

        The time the page was published. This field is absent when the publication time is not known.