Verified, then logged.
Sun 22 Nov 2026 · 00:22 KST / 15:22 UTC한국어 | ENSearch /

02 / IT · Analysis

AWS opens web search API for direct calls without a gateway

For game development pipelines that call search directly, developers manage retries, caching and error handling.

In a data center, an out-of-focus technician slides one module from the middle of a server rack divided into seven tiers to replace it. AI-generated image

AWS released a web search API that returns up to 25 results per query for developers with permission to call it. It provides direct access from code to Amazon Bedrock AgentCore’s search capability.

According to AWS, the API adds a direct route to its existing search infrastructure. Developers can use it without first setting up a gateway to handle search requests.

AWS presented custom agents and data pipelines as uses for direct access. For a game development pipeline, the choice centres on who manages search calls: the developer’s code or the gateway.

Developers’ responsibilities with direct calls

AWS described direct access as an option for developers who want to manage retries, caching and error handling themselves. The integration puts search into an existing application’s request flow.

The AWS announcement states that developers have no resources to create before calling the API. Authentication uses AWS Identity and Access Management, the permissions system used by other AWS services.

AWS recommended the existing gateway route for most agent workloads. The gateway combines search with other tools and manages authentication for incoming and outgoing requests.

Search results and their limits

According to the official documentation, responses contain ranked results and text extracts. Page titles, URLs and publication times are returned only when the search index has that information.

AWS introduced filters for domains and publication times. Applied to documentation searches for game development tools, these let developers define the search scope when making a call.

AWS also described an integration in which developers’ code checks the returned search data. Choosing direct access therefore leaves developers to design the code that processes those results. The announcement did not disclose criteria for guaranteeing how current the search results are.