Amazon Bedrock AgentCore is in preview release and is subject to change.

Creating and configuring workload identities

You can create agent identities using several methods, including the AWS CLI and the AgentCore SDK, depending on your workflow and integration requirements. AgentCore Identity provides multiple interfaces for identity creation including command-line tools for automation and scripting and programmatic APIs for integration with existing systems. Each creation method supports the full range of identity configuration options while providing appropriate interfaces for different use cases and user preferences.

Managing identities with AWS CLI

The AWS CLI provides a straightforward way to create and delete agent identities.

Create an identity:

aws bedrock-agentcore create-workload-identity \
    --name "my-agent"
            

Delete an identity:

aws bedrock-agentcore delete-workload-identity \
    --name "my-agent" \
            

Creating identities with the AgentCore SDK

The AgentCore SDK provides support for creating workload identities in Python.

Python example:

from bedrock_agentcore.services.identity import IdentityClient

# Initialize the client
identity_client = IdentityClient("us-east-1")

# Create a new workload identity for agent
response = identity_client.create_workload_identity(
    name='my-python-agent'
)

agentArn = response['workloadIdentityArn']

print(f"Created agent identity with ARN: {agentArn}")