Skip to main content

AI-assisted development with Okteto

Okteto Development Environments let AI agents deploy, test, and debug against real infrastructure instead of editing files locally. This tutorial walks you through deploying the movies sample app to Okteto, connecting Claude Code to the live environment with the Okteto plugin, and using the agent to add a feature while validating against running services.

By the end, you have:

  • A running multi-service app in your Okteto Namespace
  • An AI agent connected to your environment through the Claude Code plugin
  • A new feature developed and tested against live services

Prerequisites​

Step 1: Install the Okteto plugin for Claude Code​

The Okteto plugin gives Claude Code built-in knowledge of Okteto CLI commands and teaches it to discover your project's services from okteto.yaml.

Open Claude Code and run:

/plugin marketplace add okteto/okteto-claude-plugins
/plugin install okteto

The plugin loads automatically for any project that has an okteto.yaml file.

Step 2: Clone and deploy the movies app​

Clone the sample application:

git clone https://github.com/okteto/movies
cd movies

The movies app is a microservices application with five services:

ServiceLanguageRole
frontendReactWeb UI for browsing and renting movies
catalogNode.jsREST API serving the movie catalog from MongoDB
rentJavaAccepts rental requests and publishes to Kafka
workerGoConsumes rental events from Kafka and writes to PostgreSQL
apiGoServes rental records from PostgreSQL

Deploy the full environment:

okteto deploy --wait

This builds all service images with the Okteto Build Service, deploys the infrastructure (PostgreSQL, MongoDB, Kafka), and starts all five services. The --wait flag ensures everything is running before you proceed.

Verify the deployment:

okteto endpoints

You see URLs for the frontend and API services. Open the frontend URL in your browser to confirm the app is running.

Step 3: Start a development session​

Start a development session for the API service:

okteto up api

This activates File Sync between your local api/ directory and the Development Container running in Okteto. Code changes you or the agent make locally appear in the container within seconds.

Keep this terminal open — okteto up runs in the foreground.

Step 4: Develop with the agent​

Open a new terminal in the movies directory and start Claude Code:

claude

The Okteto plugin detects the okteto.yaml and loads context about the project's services, build targets, and test definitions automatically.

Ask the agent to add a feature:

Add a /health endpoint to the api service that returns JSON with
the server status, database connectivity, and server uptime.

The agent reads the API service code, writes the new endpoint and accompanying tests, and validates everything using okteto exec to run commands in the active Development Container:

okteto exec -- go build ./...
okteto exec -- go test ./...

Because okteto up is running, file changes sync to the container automatically. The agent builds, tests, and iterates without rebuilding container images.

note

The agent never runs okteto up — that's your job. In collaborative mode, you manage the dev session while the agent operates inside it with okteto exec. See Collaborative Workflows for details.

Step 5: Verify the changes​

After the agent finishes, test the new endpoint. Get the API service URL:

okteto endpoints

Verify the health endpoint responds:

curl -s https://<your-api-endpoint>/health | jq .

Expected output:

{
  "status": "ok",
  "database": "connected",
  "uptime": "5m12s"
}

The response comes from the live Development Container, hitting the real PostgreSQL database running in your Namespace.

Step 6: Run the tests​

With okteto up still running, ask the agent to run the tests it created:

Run the Go tests for the api service and verify the health endpoint works.

The agent runs the tests inside the Development Container:

okteto exec -- go test -v ./...

The tests execute against the live environment — the same database, same network, same configuration your app uses in production.

Step 7: Clean up​

Stop the development session by pressing Ctrl+C in the terminal running okteto up.

When you're done with the environment entirely, destroy it:

okteto destroy

Summary​

You deployed a multi-service application to Okteto, connected an AI agent to the live environment, and used it to develop and validate a feature against running infrastructure — databases, message queues, and dependent services.

The key pattern: you manage the development session with okteto up, and the agent operates inside it with okteto exec. File Sync handles code changes, so there are no rebuilds between iterations.

Next steps​