Setting Up a Strands Agent with AWS Bedrock

Overview of Strands Agents

Strands Agents is an open-source SDK from AWS that simplifies the development of AI agents through a model-driven approach. It allows you to define a task-specific prompt and tools the agent can use, and relies on foundation models like Claude, LLaMA, or models available in Amazon Bedrock.

With Strands, developers can create intelligent agents that plan, reason, and reflect while completing tasks using minimal code. It is already used internally by several AWS services such as Amazon Q Developer and AWS Glue.

Drive Business Growth with AWS's Machine Learning Solutions

Scalable
Cost-effective
User-friendly

Connect Today

Prerequisites

Before setting up your Strands Agent, ensure the following:

AWS CLI installed and configured
Active AWS account with Amazon Bedrock access
Python 3.10+
A virtual environment manager (e.g. venv)

Create Project Environment

Set up a Python virtual environment and install required packages over the terminal:

<em>python3 -m venv venv</em>

<em>source venv/bin/activate</em>

<em>pip install strands-agents strands-agents-tools    </em>

python3 -m venv venv

source venv/bin/activate

pip install strands-agents strands-agents-tools

4) Configure AWS Access

Make sure AWS credentials are configured properly using:

<em>aws configure </em>

1	<em>aws configure </em>

Build and Run a Basic Strands Agent

In agent.py, define your agent:

<em>import os</em>

<em>import logging</em>

<em># CRITICAL: Set AWS environment variables BEFORE importing strands</em>

<em>os.environ['AWS_PROFILE'] = 'profile_name'</em>

<em>os.environ['AWS_REGION'] = 'us-east-1'  </em>

<em> </em>

<em>from strands import Agent</em>

<em>from strands_tools import calculator, current_time, use_aws</em>

<em># Create the agent with tools and specify the model</em>

<em>agent = Agent(</em>

<em>    tools=[calculator, current_time,use_aws],</em>

<em>    model="anthropic.claude-3-5-sonnet-20240620-v1:0"  # Specify your model</em>

<em>)</em>

<em> </em>

<em> </em>

<em># Define a message to test the agent</em>

<em>message = """</em>

<em>I have 3 requests:</em>

import os

import logging

# CRITICAL: Set AWS environment variables BEFORE importing strands

os.environ['AWS_PROFILE'] = 'profile_name'

os.environ['AWS_REGION'] = 'us-east-1'

from strands import Agent

from strands_tools import calculator, current_time, use_aws

# Create the agent with tools and specify the model

agent = Agent(

tools=[calculator, current_time,use_aws],

model="anthropic.claude-3-5-sonnet-20240620-v1:0" # Specify your model

)

# Define a message to test the agent

message = """

I have 3 requests:

What is 7 * 8?

1

 What is 7 * 8?
What is the time right now in india?

1

 What is the time right now in india?
list count of s3 buckets in my aws account?

1

 list count of s3 buckets in my aws account?

<em>"""</em>

<em># Run the agent and handle any errors</em>

<em>if __name__ == "__main__":</em>

<em>    try:</em>

<em>        print("Running the agent...")</em>

<em>        response = agent(message)</em>

<em>        print("\nAgent Response:")</em>

<em>        print(response)</em>

<em>    except Exception as e:</em>

<em>        print(f"Error: {e}")</em>

<em>        import traceback</em>

<em>        traceback.print_exc()</em>

"""

# Run the agent and handle any errors

if __name__ == "__main__":

try:

print("Running the agent...")

response = agent(message)

print("\nAgent Response:")

print(response)

except Exception as e:

print(f"Error: {e}")

import traceback

traceback.print_exc()

Run your agent on the terminal:

<em>Python agent.py</em>

1	<em>Python agent.py</em>

Output:

Handle Common Errors

AccessDeniedException for Model ID

Make sure you’re using a region where your AWS account has access to the Bedrock model.
Double-check IAM permissions for Bedrock.

Region Configuration

Ensure AWS_REGION is exported correctly before importing the strands module.

Conditional Write Error

If you encounter an error like “pre-condition failed”, verify if you’re trying to overwrite an existing object in S3 via a conditional call — this doesn’t apply directly to Strands, but could arise if tools use S3.

Clean Up Resources

Once testing is done:

Deactivate Bedrock Model access enabled
Deactivate and remove your virtual environmentml

AWS Media Services Excellence: Unleash Potential with Cloud Consulting

Embrace Innovation!
Transform your Media Capabilities

Consult Now

About CloudThat

CloudThat is a leading provider of Cloud Training and Consulting services with a global presence in India, the USA, Asia, Europe, and Africa. Specializing in AWS, Microsoft Azure, GCP, VMware, Databricks, and more, the company serves mid-market and enterprise clients, offering comprehensive expertise in Cloud Migration, Data Platforms, DevOps, IoT, AI/ML, and more.