Copied to Clipboard
Deploying to Amazon EKS
First authenticate:
aws login --remote
Then cache the credentials locally:
xbill@penguin:~/gemini-cli-aws/adkui-eks$source save-aws-creds.sh
Exporting AWS credentials...
Successfully saved credentials to .aws_creds
The Makefile will now automatically use these for deployments.
Then start the deployment:
✦ Deployment to Amazon EKS was successful.
Deployment Summary
- EKS Cluster: adkui-eks-cluster (Status: ACTIVE)
- Image: 106059658660.dkr.ecr.us-east-1.amazonaws.com/adk-comic-image:latest
- Pod Status: Running (1/1 READY)
- Service Endpoint: http://af62eb56d13b74cefb372550e726efaa-1528063823.us-east-1.elb.amazonaws.com
The make deploy command completed the following steps:
1. Updated kubeconfig for the EKS cluster.
2. Built the Docker image based on the Dockerfile.
3. Logged in to Amazon ECR and pushed the image.
4. Generated k8s-deployment.yaml and applied it to the cluster.
You can now access the ADK Web UI at the endpoint listed above.
You can validate the final result by checking the messages:
✦ The EKS LoadBalancer endpoint is:
http://af62eb56d13b74cefb372550e726efaa-1528063823.us-east-1.elb.amazonaws.com
You can then get the endpoint:
✦ The EKS LoadBalancer endpoint is:
http://af62eb56d13b74cefb372550e726efaa-1528063823.us-east-1.elb.amazonaws.com
The service will be visible in the AWS console. The console will look similar to:
Running the ADK Web Interface
Start a connection to the EKS Deployed ADK:
http://af62eb56d13b74cefb372550e726efaa-1528063823.us-east-1.elb.amazonaws.com
This will bring up the ADK UI. Select the sub-agent "Agent3":
This will generate the Comic by using a multi-agent pipeline:
Once the multi Agent system is complete:
Visual Edit Agent Pipeline
The version of the ADK Deployed includes a visual builder:
Run the Online Viewer Agent
Once Agent3 has completed — go to the ADK agent selector and select "Agent4". This agent will allow you to browse your online comic:
View the Final Artifacts
You can use Agent4 to visualize the results of the agent pipeline:
and the final panels:
Summary
The Agent Development Kit was used to visually define a multi Agent pipeline to generate comic book style HTML. This Agent was tested locally with the CLI and then with the ADK web tool. Then, several sample ADK agents were run directly from the EKS deployment in AWS. This approach validates that cross cloud tools can be used — even with more complex agents.