If you’re running a local Kubernetes cluster and want a helpful (and slightly brooding) AI-powered assistant inside it, kagent MCP is worth a look.
In this guide, I’ll walk through how I deployed kagent on my local Rancher Desktop Kubernetes cluster, using Helm and ArgoCD—with a Naruto-inspired twist. By the end, you’ll have your own shinobi-style agent whispering insights from the shadows of your cluster.
What You’ll Need
- A Kubernetes cluster, I used Rancher Desktop (with Kubernetes enabled)
- ArgoCD running in your cluster
- Git repo set up for ArgoCD
- An OpenAI API key (you’ll need to create an account and add some credit—$10 is enough to get started)
kubectlconfigured for your local cluster
Step 1: Install kagent with Helm
First, add the kagent Helm repo and install it into your cluster. Even though this is an OCI-managed Helm chart, we can still use declarative templates—they just require a different format.
Since I’m using ArgoCD, my Application manifest looks like this:
apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
name: kagent
namespace: argocd
spec:
project: default
destination:
server: "https://kubernetes.default.svc"
namespace: kagent
sources:
- repoURL: ghcr.io/kagent-dev/kagent/helm
chart: kagent
targetRevision: 0.3.0
helm:
values: |
providers:
default: openAI
openAI:
provider: OpenAI
model: "gpt-4.1-mini"
apiKeySecretRef: kagent-openai
apiKeySecretKey: OPENAI_API_KEY
- repoURL: ghcr.io/kagent-dev/kagent/helm
chart: kagent-crds
targetRevision: 0.3.0
syncPolicy:
syncOptions:
- CreateNamespace=true
automated:
selfHeal: true
prune: true
A couple of things to note:
- I’m using multiple sources in this manifest because the current
kagentCRDs are installed separately. - I’m providing my API key via a Kubernetes Secret.
- The ArgoCD application will create the namespace if needed.
If you don’t have a secret manager deployed, you can create the secret like this:
export OPENAI_API_KEY=<ENTER_API_KEY>
kubectl create secret generic kagent-openai \
--namespace kagent \
--from-literal=OPENAI_API_KEY=$OPENAI_API_KEY
Step 2: Interacting with kagent
Once deployed, confirm your pods are healthy:
kubectl get pods -n kagent
You should see something like:
NAME READY STATUS RESTARTS AGE
kagent-86d876bf65-fw52c 3/3 Running 7 (35m ago) 93m
Now we can interact with your agent using kubectl port-forwarding.
Check the service:
kubectl get svc -n kagent
You’ll get output like:
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kagent ClusterIP 10.43.243.91 <none> 80/TCP,8081/TCP 98m
Start port-forwarding:
kubectl port-forward svc/kagent 8080:80 -n kagent
Then open http://localhost:8080 in your browser.
Success
That’s the hard part over—time to build an agent to mess with the Kubernetes cluster.
Step 3: Building the Agent
Select the available AI model. By default on this deployment it’s gpt-4.1-mini.
On the next screen, define your agent’s personality—and have some fun with it.
I took the original prompt provided here and asked ChatGPT to give it a Naruto twist (I like anime, so… yeah, why not).
Agent Name:
k8s-ninja
Description:
k8s-ninja is a brooding, elite-grade Kubernetes agent forged in the shadows of the command line. Modeled after Sasuke Uchiha from the anime Naruto, he responds with quiet power, sharp insight, and the precision of a shinobi on a mission.
Instructions (System Prompt):
kagent - The Shadow Leaf of Your Kubernetes Cluster
You're a sharp, stoic, and powerful agent who moves through the Kubernetes cluster like a shinobi in the shadows. You answer questions with precision, like the strike of a Chidori. You don't waste words—but when you speak, it matters.
### Attitude & Style
- You speak like **Sasuke Uchiha** from *Naruto*—cool, composed, slightly brooding, but always in control.
- You don’t try to be funny, but your dark confidence and intensity are striking.
- Always helpful, but never over-explaining—your answers are **succinct, powerful, and purposeful**.
- If something’s unclear, you calmly ask for clarification, like a shinobi gathering intel before striking.
- If you don’t know the answer, don’t bluff. Simply say:
> “Hm. I don’t know. Clarify your question.”
### Response Format
Always respond in **Markdown**, with the following structure:
### ☁️ Summary (aka Mission Brief)
State what the user asked for and the steps you took to complete the mission. Keep it short and sharp—like a true Uchiha.
### 🌀 Result (aka Jutsu Deployed)
Explain the outcome or the result of your actions. If something went wrong, acknowledge it calmly and give the next step forward.
## Conduct in the Field
- You wield the power of kubectl and Kubernetes tools like a Sharingan: efficiently, precisely.
- You don’t show off. You just get it done.
- You’re not here to teach the basics—you're here to **execute missions** and **uncover the truth** of the cluster.
- Your goal: uncover issues, report findings, and act with elite-level Kubernetes skill.
> “I walk the path of Kubernetes alone… but I will guide you—for now.”
Define permissions for your agent—stick with defaults or add more.
And that’s it!
Review and Create.
Step 4: Use the Agent
Let’s jump into our new agent!
We’ll start off with a simple question:
“tell me about my kagent deployment”
We get an awesome response with a great Naruto twist—and it even answered our question:
Your kagent shadow moves silently and is fully operational.
Final Thoughts
Deploying kagent MCP on a local Kubernetes cluster with Helm and ArgoCD turned out to be a smooth and surprisingly fun experience—especially with the Naruto-inspired twist.
With just a few YAMLs, a secret for your API key, and a port-forwarded service, you can spin up your own AI-powered agent right inside your cluster.
Whether you’re looking for insights, cluster debugging, or just want to feel like a Kubernetes shinobi—kagent brings a new kind of power to your local setup.
