Strands CLI Tool

A command-line utility for generating and deploying Strands agents to Amazon EKS.

Features

• Create a standardized project structure for Python-based Strands agents
• Generate boilerplate code for Strands agents with FastAPI wrappers
• Produce Docker configurations for containerization
• Generate Kubernetes manifests and/or Helm charts for EKS deployment
• Provide deployment guidance and scripts for various environments

Installation

pip install strands-cli

For development:

git clone https://github.com/yourusername/strands-cli.git
cd strands-cli
pip install -e ".[dev]"

Usage

Initialize a New Strands Agent Project

strands-cli init my-weather-agent --description "Weather forecast agent"

This command creates a project skeleton with the core directories and files needed for agent development (agent code, API wrapper, Docker configuration). It does not create any deployment directories or files. Deployment assets for Helm and Kubernetes are only generated when you explicitly run the respective generate commands.

Build Docker Image

cd my-weather-agent
strands-cli build --push

Run Locally with Streamlit UI

Test your agent locally with an interactive chat UI before deploying:

cd my-weather-agent
strands-cli run

This will:

• Build your agent image (if needed)
• Start your agent in a container
• Launch a Streamlit UI for interactive testing

You can access the UI at http://localhost:8501 and the agent API at http://localhost:8000.

Important for Docker Desktop users (macOS/Windows): You need to enable file sharing for your home directory to allow AWS credentials access. Go to Docker Desktop → Settings → Resources → File Sharing and add your home directory (e.g., /Users/username) to the list of shared folders. The agent container will mount your local .aws directory to /home/appuser/.aws inside the container.

Options:

# Custom ports
strands-cli run --port 3000 --agent-port 8080

# Run without UI (headless mode)
strands-cli run --no-ui

# Run in background
strands-cli run --detach

# Use an existing image instead of building locally
strands-cli run --image-uri registry/name:tag

# Specify AWS profile for credentials
strands-cli run --aws-profile myprofile

Generate Deployment Assets

After building your Docker image, you need to explicitly generate all deployment assets:

Generate Helm Chart

# Using set values
strands-cli generate helm --set image.repository=my-registry/my-weather-agent --set image.tag=latest

# Using direct flags (recommended)
strands-cli generate helm --image-uri my-registry/my-weather-agent --image-tag latest --service-account my-bedrock-sa

This command generates all necessary Helm files, including:

• Chart.yaml file
• Helm template files (deployment.yaml, service.yaml, etc.)
• values.yaml with your configuration
• VALUES.md documentation

Generate Raw Kubernetes Manifests

# Basic usage
strands-cli generate k8s --namespace my-agents

# With image and service account configuration
strands-cli generate k8s --namespace my-agents \
  --image-uri my-registry/my-weather-agent \
  --image-tag latest \
  --service-account my-bedrock-sa

Note: All deployment assets are only generated when you explicitly run the generate commands. This allows you to include specific parameters like image URIs and service account names.

Create Pod Identity for AWS Bedrock Access

strands-cli create-pod-identity my-bedrock-sa arn:aws:iam::aws:policy/AmazonBedrockFullAccess --namespace my-agents

Documentation

For detailed documentation, see the SPECIFICATIONS.md file.

License

MIT