AWS launches direct Web Search API for Bedrock AgentCore

AWS has introduced the Web Search API on Amazon Bedrock AgentCore, allowing applications to retrieve ranked web results through a single API call. The service is designed for developers who need fresh web information without building an AgentCore Gateway first.
- AWS introduced the Web Search API on Amazon Bedrock AgentCore.
- Applications can call InvokeWebSearch through AWS SDKs or the AWS CLI.
- Results come from a web index operated by Amazon and include text extracts.
- Developers can filter by domain and publication time.
- The API uses IAM credentials and requires no resource to be created first.
Direct access to Amazon's web search index
The Web Search API returns ranked results from a web index operated by Amazon. Each result includes a text extract and, when available, the page title, URL and publication time. Queries are served inside AWS and are not sent to a third-party search engine, according to AWS.
Developers can call the InvokeWebSearch operation through AWS SDKs or the AWS Command Line Interface. Requests use AWS Signature Version 4 and the same AWS Identity and Access Management credentials used for other AWS services. No resource needs to be created before making a request, and the API does not require a separate API key or secret.
Built for applications beyond agent gateways
AWS says AgentCore Gateway remains the recommended path for most agentic workloads. The gateway provides a secure Model Context Protocol endpoint that can combine web search with other tools, while also handling authentication, semantic tool selection and optional MCP sessions.
The standalone API targets developers who need more direct control. AWS lists custom agent loops, orchestration frameworks such as LangChain and LlamaIndex, search-powered dashboards, fact-checking pipelines, content enrichment workflows and monitoring systems as potential use cases.
It also allows developers to manage retries, caching and error handling in their own code. Applications with an existing API layer can add web grounding without adopting the full AgentCore Gateway stack. The API uses the same search infrastructure that powers Alexa+, Kiro, Quick and other Amazon products, with the same quality and compliance posture, according to AWS.
Filtering, permissions and result controls
The API supports domain filters that include or exclude specific websites. Each list can contain up to 100 domain names, and a root domain also matches its subdomains. Developers can also restrict results by publication time using publishedAfter and publishedBefore RFC 3339 date-times in UTC, with both bounds treated as inclusive.
Each query can request between 1 and 25 results. If maxResults is not set, the API returns up to 25. Search text can contain up to 200 characters, and every result includes a text extract that can be passed into a model's context window.
Access requires the bedrock-agentcore:InvokeWebSearch IAM permission on the AWS managed web search tool. AWS says the same action and resource are used by a gateway's service role, allowing one policy to cover both access routes. Calls from AWS Lambda, Amazon Elastic Container Service, Amazon Elastic Compute Cloud and Amazon Elastic Kubernetes Service can use credentials supplied by the execution role.
What the API means for developers
The launch gives developers a lower-level way to add current web information to software that does not need the rest of AgentCore Gateway. That can reduce the integration requirements for existing applications and pipelines, while retaining AWS authentication and IAM-based access controls.
AWS also documents integration with agent frameworks. LangChain agents can access the service through the AgentCore web search tool in langchain-aws 1.9.3 and later. The tool can enforce rules such as excluded domains, a publication date floor and a search budget, while returning results as data that application code can check.


