View a markdown version of this page

Web Search Tool - Amazon Bedrock AgentCore
Key capabilitiesPurpose-built web indexPrivate by designHow it worksUse casesSet up Web Search ToolConfigure domain filteringConfigure the Gateway Service RoleInput schemaResponse formatAvailabilityAcceptable use

Web Search Tool

AI agents are transforming how organizations interact with information, but they face a fundamental limitation: their knowledge is frozen at training time. When a user asks about today’s stock prices or a software release that shipped an hour ago, an agent relying solely on its training data cannot provide accurate answers.

Building custom web search integrations to solve this is costly — procuring third-party search APIs, managing keys and quotas and rate limits, parsing inconsistent result formats across providers, building snippet extraction logic so models get relevant passages instead of raw HTML, reasoning about where customer queries travel and how data might be retained, and maintaining freshness and coverage over time. Each of these is a project in itself.

Web Search Tool on Amazon Bedrock AgentCore eliminates that complexity. It is a fully managed, MCP-compliant web search capability that lets your agents retrieve information from the web without any infrastructure overhead. It is available as a managed connector that you connect to your AgentCore Gateway. Agents discover it with a standard tools/list call and invoke it like any other Model Context Protocol tool. There are no search APIs to provision, no outbound credentials to manage, and no result-parsing glue to maintain.

The following sections walk through the key capabilities, the architecture of the purpose-built web index, privacy guarantees, use cases, and how to set up Web Search Tool.

Key capabilities

  • Up-to-date information access — Retrieve current web results with titles, URLs, snippets, and publication dates to ground agent responses in up-to-date information.

  • Zero infrastructure management — No third-party search APIs to provision or scaling to configure. The Gateway exposes web search as a standard MCP tool automatically.

  • Framework agnostic — Works with Strands Agents, LangChain, LangGraph, CrewAI, or any MCP-compatible client through the standard Model Context Protocol.

  • Purpose-built web index — Backed by an Amazon-operated web index spanning tens of billions of documents.

  • Knowledge graph for high-confidence facts — Grounds entities and their relationships to provide factual answers rather than leaving the model to infer them from extracted page text.

  • Semantic snippet extraction — Pulls semantically relevant passages from each web page, optimized for a model’s context window, rather than returning raw HTML or full pages.

Purpose-built web index

Many solutions that add web search to an agent are wrappers over a third-party search engine. Web Search Tool on Amazon Bedrock AgentCore is different — it is backed by a web index that Amazon builds and operates directly.

Broad coverage

The index spans tens of billions of documents. That scale matters for coverage — long-tail questions about niche libraries or obscure product specifications can be more effectively answered when the index is broad rather than limited to the most popular pages.

Continuously updated

Amazon refreshes the index on an ongoing basis, reflecting new and changed content within minutes. For agents answering questions about price movements, or just-published announcements, that recency window is the difference between a grounded answer and a confidently wrong one. When your agent searches for "what happened today," the results reflect what actually happened today.

Knowledge graph for high-confidence facts

Web Search Tool includes a built-in knowledge graph that grounds entities and their relationships. For factual questions — such as who holds a role or when something was founded — the knowledge graph provides high-confidence answers rather than leaving the model to infer them from extracted page text. This reduces the kind of subtle factual drift that can occur when an agent stitches together an answer from snippets alone.

Semantic snippet extraction optimized for model context

Rather than handing the model a raw HTML dump or a full page, the tool performs semantically relevant snippet extraction. It pulls the passages from each web page that actually bear on the query and returns them in a form optimized for a model’s context window. The result is denser, more relevant context per token — the model sees the parts that matter and spends fewer tokens on boilerplate and navigation chrome, which helps improve the precision of cited answers.

Private by design

For many enterprises, the question that stalls a web search rollout isn’t "does it work" — it’s "where do my users' queries go, and what happens to them?" Web Search Tool is built so those answers are straightforward.

Queries never leave AWS

When your agent issues a search, the query is served entirely within AWS infrastructure. Customer queries are not sent to a third-party search engine or routed outside AWS. The Gateway authenticates to the AWS-owned connector and routes the request internally, so the data path stays inside AWS end to end. For teams with data egress concerns, this removes an entire category of review.

How it works

The Web Search Tool is a built-in connector. The Gateway handles schema management, parameter governance, endpoint resolution, and service authentication.

  1. Gateway setup — Create a Gateway and add a Web Search Tool target using connectorId: "web-search". The Gateway snapshots the tool schema and provisions the integration.

  2. Tool discovery — Your agent calls tools/list on the Gateway endpoint and discovers the WebSearch tool with its input schema.

  3. Search invocation — Your agent calls tools/call with a natural language query (up to 200 characters). The Gateway authenticates to the backend and routes the request internally within AWS.

  4. Structured results — The tool returns results with semantically relevant text snippets, URLs, titles, and publication dates in both text and structured JSON formats.

  5. Grounded response — Your agent uses the results to compose a response with cited sources.

Use cases

  • Up-to-date research — Answer questions about current events, stock prices, or recent announcements by searching the web and synthesizing results with source citations.

  • Fact-checking — Verify claims in documents by searching for primary sources and confirming or correcting figures using knowledge graph-backed factual answers.

  • Competitive intelligence — Monitor competitor announcements, product launches, or pricing changes from public web sources.

  • Stale knowledge elimination — Give agents access to information published after their training cutoff, such as new software releases, policy changes, or breaking developments.

Set up Web Search Tool

For instructions on how to create a Gateway Target with the Web Search Tool connector configuration, including setup examples using the Python SDK and CLI, see Set up Web Search Tool in the target configuration guide.

Configure domain filtering

You can restrict which domains the Web Search Tool is allowed to query by configuring a domain denylist. For configuration examples and details, see Configure domain filtering in the target configuration guide.

Configure the Gateway Service Role

The Gateway needs a service role that allows the AgentCore service to perform actions on your behalf. For the required IAM permissions and policy configuration, see Configure the Gateway Service Role in the target configuration guide.

Input schema

The Web Search Tool accepts the following input schema when invoked via tools/call:

{
  "inputSchema": {
    "type": "object",
    "properties": {
      "query": {
        "description": "The search query string",
        "type": "string"
      },
      "maxResults": {
        "description": "Maximum number of results to return. Valid range: 1-25. Defaults to 10.",
        "type": "integer"
      }
    },
    "required": ["query"]
  }
}
Field Type Required Description

query

string

Yes

The search query string. Must be 200 characters or fewer.

maxResults

integer

No

Maximum number of results to return. Valid range: 1-25. Defaults to 10.

Response format

The Web Search Tool returns results in the following MCP-compliant format:

{
  "isError": false,
  "content": [
    {
      "type": "text",
      "text": "{\"id\":\"824f89d0\",\"results\":[{\"text\":\"Python 3.13 was released on October 7, 2024, featuring improvements to the interactive interpreter, experimental free-threaded mode, and a preliminary JIT compiler...\",\"publishedDate\":\"2024-10-07\",\"url\":\"https://example.com/python/releases/3.13\",\"title\":\"Python 3.13 Release Highlights\"}]}"
    }
  ]
}
Field Type Required Description

text

string

Yes

Text content or snippet of the search result

url

string

No

URL of the source webpage

title

string

No

Title of the source webpage

publishedDate

string

No

Publication date of the result

Availability

The Web Search Tool connector is available in the US East (N. Virginia) us-east-1 Region.

Because Amazon operates the full search stack, improvements to freshness, coverage, relevance, and snippet quality flow to your agents automatically through the same managed connector. No version upgrades or migrations are needed on your side.

Acceptable use

If you use Web Search Tool, you are responsible for your use, and any use by your end users, of content retrieved from Web Search Tool (“Search Results”). You must retain and display the source citations and links provided with each Search Result in any output you surface to your end users that uses the Search Result. You may not use Web Search Tool to (a) extract, store, or reproduce content from Search Results in bulk, or (b) build or populate a competing index or database.