Skip to content

README.md

Latest commit

 

History

History
117 lines (81 loc) · 3.59 KB

README.md

File metadata and controls

117 lines (81 loc) · 3.59 KB

AWS CDK Multi-Modal Agent Deployment

Introduction

This is a Python-based CDK (Cloud Development Kit) project that demonstrates how to deploy serverless Lambda functions implementing the Strands Agent framework. The project includes two Lambda functions:

  1. Weather Forecasting Agent - A simple agent that provides weather forecasting capabilities
  2. Multi-Modal Processing Agent - An advanced agent that can process and analyze different types of media (images, documents, videos)

Prerequisites

  • AWS CLI installed and configured
  • Python 3.8 or later
  • jq (optional) for formatting JSON output
  • AWS account with Bedrock access

Project Structure

  • agent_lambda/ - Contains the CDK stack definition in Python
  • app.py - Main CDK application entry point
  • layers/ - Contains Lambda layers for the Strands Agent framework
    • package_for_lambda.py - Python script that packages Lambda code and dependencies into deployment archives
    • lambda_requirements.txt - Dependencies for the Lambda functions
  • lambdas/code/ - Contains the Lambda function code
    • lambda-s-agent/ - Weather forecasting agent Lambda function
    • lambda-s-multimodal/ - Multi-modal processing agent Lambda function

Setup and Deployment

  1. Create a Python virtual environment and install dependencies:
# Create a Python virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install CDK dependencies
pip install -r requirements.txt

# Install Python dependencies for lambda with correct architecture
pip install -r layers/lambda_requirements.txt --python-version 3.12 --platform manylinux2014_aarch64 --target layers/strands/_dependencies --only-binary=:all:
  1. Package the lambda layers:
python layers/package_for_lambda.py
  1. Bootstrap your AWS environment (if not already done):
cdk bootstrap
  1. Deploy the stack:
cdk deploy

Usage

After deployment, you can invoke the Lambda functions using the AWS CLI or AWS Console.

Weather Forecasting Agent

aws lambda invoke --function-name AgentSFunction \
      --region us-east-2 \
      --cli-binary-format raw-in-base64-out \
      --payload '{"prompt": "What is the weather in New York?"}' \
      output.json

Multi-Modal Processing Agent

aws lambda invoke --function-name MultimodalSFunction \
      --region us-east-2 \
      --cli-binary-format raw-in-base64-out \
      --payload '{"prompt": "Analyze this image", "s3object": "s3://your-bucket/path/to/image.jpg"}' \
      output.json

If you have jq installed, you can output the response from output.json like so:

jq -r '.' ./output.json

Otherwise, open output.json to view the result.

Multi-Modal Processing Capabilities

The Multi-Modal Processing Agent can handle various types of media:

  • Images: PNG, JPEG/JPG, GIF, WebP
  • Documents: PDF, CSV, DOCX, XLS, XLSX
  • Videos: MP4, MOV, AVI, MKV, WebM

The agent uses custom tools built with the Strands Agent framework to process and analyze these media types.

Cleanup

To remove all resources created by this example:

cdk destroy

Additional Resources