Deploying Kagent on AWS EKS: Setup Guide and LLM Selection Strategy

Transform Kubernetes Operations with Intelligent AI Agents A practical guide to deploying Kagent on Amazon EKS, integrating Large Language Models (LLMs), and building AI-powered agents for Kubernetes troubleshooting, observability, automation, and platform engineering.

Sandipkumar Patel

Published Sep 15, 2026

Last modified Sep 15, 2026

Introduction

Cloud-native environments continue to grow in complexity. Managing Kubernetes clusters requires engineers to analyze logs, inspect resources, troubleshoot failures, and automate operational tasks. Kagent simplifies these challenges by introducing AI-powered agents capable of interacting with Kubernetes using natural language. Kagent as an open-source framework that enables AI agents to operate within Kubernetes environments and assist with cluster management.

This guide demonstrates how to deploy Kagent on AWS EKS and how to choose the right Large Language Model (LLM) for your use case.

In this article, you will learn:

- How to deploy Kagent on Amazon EKS
- How Kagent works with AI models
- Which LLMs are supported
- GPT-5.3 Codex vs GPT-5.6 Sol vs GPT-5.6 Luna
- Which model is best for different Kagent use cases
- Recommended enterprise deployment architectures
- Best practices for AI-powered Kubernetes operations

Introduction to Kagent

Kagent is an agentic AI framework designed specifically for Kubernetes environments. Rather than manually executing multiple kubectl commands and interpreting raw output, users can interact with intelligent agents using natural language.

Why Deploy Kagent on AWS EKS?

Amazon Elastic Kubernetes Service (EKS) provides a fully managed Kubernetes platform that simplifies cluster deployment and management.

Combining Amazon EKS with Kagent enables:

- Natural language Kubernetes operations
- Faster troubleshooting
- AI-assisted incident response
- Automated health assessments
- Platform engineering automation
- Enhanced observability

Benefits include simplified cluster management, faster operational workflows, and AI-powered insights for cloud-native workloads.

Instead of manually running multiple kubectl commands, users can ask questions such as

- Why is my pod failing?
- What resources are consuming the most memory?
- Show ingress configuration issues.
- Generate a cluster health report.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
                   User

                     │

                     ▼

              Kagent Dashboard

                     │

                     ▼

               Kagent Agents

                     │

       ┌───────────  ─┼──────────────┐

       │             │             │

       ▼             ▼             ▼



 Kubernetes     Cloud Tools      Logs



                     │

                     ▼



                    LLM



   GPT-5.3 Codex | Sol | Luna



                     │

                     ▼



                 AWS EKS

Hands-On Lab and Demo Repository

The full deployment walkthrough, manifests, configuration examples, and workshop resources are available in the GitHub repository:

Repository: Click Here for GitHub Repo

(https://github.com/sandip9334/Deploying-Kagent-on-AWS-EKS )

Suggested workshop flow:

1. Deploy EKS
2. Configure kubectl
3. Install Kagent
4. Configure LLM provider
5. Deploy AI agents
6. Execute operational scenarios
7. Perform Kubernetes troubleshooting
8. Review tool-calling workflows

Now Let's Try Kagent

Prerequisites

Before deploying Kagent:

- AWS Account
- Amazon EKS Cluster
- kubectl
- Helm
- OpenAI API Key or equivalent provider key

Step 1: Create an EKS Cluster

Create an EKS cluster using:

- AWS Console
- AWS CLI
- eksctl

Configure kubectl access:

1
2
3
aws eks update-kubeconfig \

  --region us-east-1 \

  --name my-eks-cluster

Validate connectivity:

1
2
3
4
5
6
7
8
9
10
11
kubectl get nodes

```【1-3a8e0e】



---



# Step 2: Configure Your AI Provider



Export the OpenAI API key:



```bash

export OPENAI_API_KEY="your-api-key"

Step 2: Configure Your LLM Provider

Export the OpenAI API Key:

1
export OPENAI_API_KEY="your-api-key"

Kagent uses the configured model provider for decision-making and tool execution.

Step 3: Install Kagent CLI

1
2
3
4
5
6
7
macOS

brew install kagent

Linux

curl https://raw.githubusercontent.com/kagent-dev/kagent/refs/heads/main/scripts/get-kagent | bash

Step 4: Install Kagent

Demo Profile

1
kagent install --profile demo

Minimal Profile

1
kagent install --profile minimal

The demo profile includes preconfigured agents and tools to accelerate learning and demonstrations.

Step 5: Open the Dashboard

1
kagent dashboard

Access:

1
http://localhost:8082

Understanding LLM Selection in Kagent

A common misconception is that the most intelligent model automatically provides the best Kagent experience. In reality, the best model is the one that:

- Uses tools effectively
- Troubleshoots Kubernetes correctly
- Understands AWS architecture
- Maintains context throughout investigations

Kagent supports multiple LLM providers, including OpenAI, Azure OpenAI, Anthropic Claude, Gemini, and Ollama.

Model Comparison for Kagent

Capability	GPT-5.3 Codex	GPT-5.6 Sol	GPT-5.6 Luna
Kubernetes Troubleshooting	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Tool Calling	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐
AWS Operations	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Architecture Reviews	⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐
Automation	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐
Reasoning Depth	⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Cost Efficiency	⭐⭐⭐	⭐⭐⭐	⭐⭐⭐⭐⭐
Speed	⭐⭐⭐⭐	⭐⭐⭐	⭐⭐⭐⭐⭐

Sample Demo Prompts

Infrastructure

List all pods in my EKS cluster.

Troubleshooting

Analyze recent Kubernetes events.

Cost Optimization

Recommend resource sizing improvements.

Observability

Generate a weekly operational summary.

Best Practices

1. Use GPT-5.3 Codex for operational troubleshooting.
2. Use Sol for architecture reviews.
3. Use Luna for repetitive reporting.
4. Enable RBAC controls.
5. Limit production permissions.
6. Test agents in non-production environments first.
7. Monitor agent tool usage.

Conclusion

Kagent represents a significant step toward AI-powered cloud operations. By combining Amazon EKS, Kubernetes tooling, and advanced language models, organizations can dramatically simplify troubleshooting, increase operational efficiency, and improve platform reliability.

For most AWS EKS deployments, GPT-5.3 Codex provides the best overall experience due to its strong tool-calling, Kubernetes understanding, and infrastructure automation capabilities. For enterprise architecture reviews, GPT-5.6 Sol excels, while GPT-5.6 Luna is ideal for reporting and high-volume automation workloads.

Some Observations and Analysis

GPT-5.3 Codex: Best Overall Choice for Kagent

Best For

- Kubernetes troubleshooting
- Amazon EKS operations
- Helm troubleshooting
- Terraform analysis
- YAML generation
- DevOps automation
- Incident response

Why It Excels

Kagent frequently performs:

- Tool execution
- Cluster inspections
- Log analysis
- Resource discovery
- Multi-step diagnostics

These workflows align closely with GPT-5.3 Codex's strengths in coding, tool usage, and infrastructure operations. The document specifically recommends GPT-5.3 Codex as the strongest option for most Kagent deployments.

Recommended Users

- Platform Engineers
- Kubernetes Administrators
- DevOps Engineers
- SRE Teams
- Skills Lab Workshops

GPT-5.6 Sol: Best for Enterprise Architecture

Best For

- Cloud architecture reviews
- Security assessments
- Governance analysis
- Cross-account AWS reviews
- Strategic recommendations
- Executive reporting

Why It Excels

Sol is better suited when Kagent acts as:

- Cloud Architect
- Platform Consultant
- Security Advisor
- Operations Strategist

Recommended Users

- Platform Architects
- Cloud Consultants
- Enterprise Operations Teams
- Security Teams

GPT-5.6 Luna: Best for Cost Optimization and Scale

Best For

- Alert summaries
- Daily reports
- Monitoring
- Health checks
- Event reviews
- Automated ticket analysis

Why It Excels

When a cluster produces thousands of events daily, Luna can handle repetitive tasks at lower cost while maintaining acceptable quality.

Example Prompts

- Summarize today's Kubernetes events.
- Generate a weekly EKS operational report.
- Create a cluster health summary.

Recommended Users

- Operations Teams
- NOC Teams
- Managed Services Teams
- High-volume environments

# devops # kubernetes # aws-user-groups # agents # amazon-eks

Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.

Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article

Builder Center

Deploying Kagent on AWS EKS: Setup Guide and LLM Selection Strategy

Sandipkumar Patel

Introduction

Introduction to Kagent

Why Deploy Kagent on AWS EKS?

Hands-On Lab and Demo Repository

Repository: Click Here for GitHub Repo

Now Let's Try Kagent

Prerequisites

Step 1: Create an EKS Cluster

Step 2: Configure Your LLM Provider

Step 3: Install Kagent CLI

Step 4: Install Kagent

Step 5: Open the Dashboard

Understanding LLM Selection in Kagent

Model Comparison for Kagent

Sample Demo Prompts

Infrastructure

Troubleshooting

Cost Optimization

Observability

Best Practices

Conclusion

GPT-5.3 Codex: Best Overall Choice for Kagent

Best For

Why It Excels

Recommended Users

GPT-5.6 Sol: Best for Enterprise Architecture

Best For

Why It Excels

Recommended Users

GPT-5.6 Luna: Best for Cost Optimization and Scale

Best For

Why It Excels

Example Prompts

Recommended Users