> SpecStream is the live web data engine for teams and agents. This is the Markdown copy of https://specstream.ai/docs/integrations/amazon-bedrock-agentcore. Every page, with what SpecStream does and costs: https://specstream.ai/llms.txt

# Amazon Bedrock AgentCore

Give the agents behind an Amazon Bedrock AgentCore Gateway live web data. SpecStream joins the gateway as an MCP server target, and agents ask for data in plain words, see the price before it runs, and get each value with the page it came from.

[SpecStream for Amazon Bedrock AgentCoreAgent frameworks & builders](https://specstream.ai/integrations/amazon-bedrock-agentcore.md)

## What you need

-   An AgentCore Gateway in a Region where AgentCore is available, with inbound auth set up.
-   A SpecStream account. New accounts get $10 of work free, with no card.
-   A SpecStream API key, which step 1 creates.

1.  1

    ## 1. Create a SpecStream API key

    Make the key in SpecStream under **Agents & API**, then **API keys**, then **Create API key**. To ask for data, the key needs **Read saved requests**, **Create and edit requests**, **Prepare work and follow-ups**, **Authorize paid work**, **Read results and their sources**, and **Read account and balance**. With **Authorize paid work** on, set a **Total spending limit** to cap what the key can spend, and set your account's own limit in **Settings**, under **Usage and billing**. Copy the key when it's shown. It isn't shown again.

2.  2

    ## 2. Store the key in AgentCore Identity

    In the AgentCore console, open **Identity** and add an API key with the SpecStream key as its value. Or create the credential provider from the AWS CLI, and note the provider ARN it returns:

    Terminal

    ```bash
    aws bedrock-agentcore-control create-api-key-credential-provider \
      --name specstream \
      --api-key <your SpecStream API key>
    ```

3.  3

    ## 3. Add SpecStream as a target

    In the AgentCore console, choose **Gateways**, select your gateway, and in **Targets** choose **Add**. Name the target `specstream`, set **Target type** to **MCP server**, and enter `https://api.specstream.ai/mcp` as the endpoint.

    For outbound authorization, choose the API key and the credential provider from step 2. Send it in the header `Authorization` with the prefix `Bearer`, then choose **Add target**. From the CLI:

    Terminal

    ```bash
    aws bedrock-agentcore-control create-gateway-target \
      --gateway-identifier <gateway ID> \
      --name specstream \
      --target-configuration '{"mcp":{"mcpServer":{"endpoint":"https://api.specstream.ai/mcp"}}}' \
      --credential-provider-configurations '[{"credentialProviderType":"API_KEY","credentialProvider":{"apiKeyCredentialProvider":{"providerArn":"<provider ARN>","credentialLocation":"HEADER","credentialParameterName":"Authorization","credentialPrefix":"Bearer"}}}]'
    ```

4.  4

    ## 4. Call SpecStream through the gateway

    When the target is created, the gateway lists SpecStream's tools and names each one after the target, such as `specstream___compile`. Agents reach them at the gateway's MCP URL with the gateway's own inbound auth.

    A request runs in four calls: `compile` reads it, `quote_capacity_work` prices it, `authorize_capacity_work` approves the price, and `list_find_rows` reads the results with their sources. Have the agent show the quote to a person before it authorizes, unless you've decided the key's limit is approval enough.

## When SpecStream adds tools

The gateway keeps the tool list it read when the target was created. Run `SynchronizeGatewayTargets` on the gateway to pick up tools SpecStream adds later.

## One account per key

Every call through the target runs on the SpecStream account that owns the key, inside the key's limits. Give each team or workload its own target and key, each with its own spending limit and only the operations it needs.

## Cost

SpecStream bills your balance per match, never above the maximum an authorization approved, and a request that finds nothing is free. AWS bills the gateway separately. [Balance, quotes, and billing](https://specstream.ai/docs/billing.md) has the rates.

## Remove

Delete the target from the gateway, then revoke the key on the **API keys** tab of **Agents & API** in SpecStream. Your saved results stay put. Questions go to [support@specstream.ai](mailto:support@specstream.ai).

## More

[SpecStream for Amazon Bedrock AgentCore](https://specstream.ai/integrations/amazon-bedrock-agentcore.md) covers what it does in Amazon Bedrock AgentCore on one page. Every other tool has its guide in [Connect your tools](https://specstream.ai/docs/integrations.md). You see the most a request can cost before it starts, and the matches are paid from your balance. [Billing](https://specstream.ai/docs/billing.md) has the details. For help, email [support@specstream.ai](mailto:support@specstream.ai).