AWS Strands

AWS customers need agents that call Lambda, S3, DynamoDB, and Bedrock models with IAM-scoped permissions, not generic tools bolted on later.

Strands is AWS's native agent wiring — plugs directly into the cloud's power outlets (IAM, Bedrock, Lambda) instead of extension cords from other clouds.

Scenario

Ops agent on Strands: reads CloudWatch alarms via tool, drafts runbook steps, invokes Lambda remediation function after human approval.

Solution

In Agent Frameworks, apply AWS Strands to this scenario: Ops agent on Strands: reads CloudWatch alarms via tool, drafts runbook steps, invokes Lambda remediation function after human approval. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to AWS Strands (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where AWS Strands happens in the code.

1from strands import Agent  # import dependencies2from strands.tools import tool  # import dependencies3
4@tool5def get_alarms(state: str) -> list:  # define a reusable function6    # calls CloudWatch API7    return [{"name": "HighCPU", "status": "ALARM"}]  # return the result8
9agent = Agent(model="bedrock:anthropic.claude-3-5-sonnet",10    system_prompt="You are an SRE assistant.", tools=[get_alarms])11agent("What alarms are firing in us-east-1?")

• Treating AWS Strands as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for aws strands

Quick recap — the most important points from this module.