AWS Builder Center
Your AI Coding Agent Knows Code. But Does It Really Know AWS?

Your AI Coding Agent Knows Code. But Does It Really Know AWS?

AI coding agents are becoming part of normal development workflows. They can generate APIs, create infrastructure files, debug applications and even help deploy projects. But there is a problem when we ask them to work with AWS. A coding agent may understand AWS in general, but that does not always mean it knows the latest service capabilities, the right API to use or the recommended AWS configuration for a particular task. This is exactly where the Agent Toolkit for AWS becomes interesting.

What is the Agent Toolkit for AWS?
The Agent Toolkit for AWS gives AI coding agents AWS-specific tools, skills and guidance.
Instead of relying only on what the model already knows, a coding agent can use AWS-provided capabilities while helping you build and manage cloud applications.
It includes three important pieces:
  • AWS MCP Server for interacting with AWS services
  • AWS Skills that provide task-specific AWS guidance
  • Plugins that combine useful tools and skills for different workflows
The interesting part is that you do not need to replace your existing coding agent.
The toolkit works with tools developers may already be using, including Kiro, Cursor, Claude Code and Codex.

Setup is surprisingly simple

If you have AWS CLI version 2.35.0 or later, you can start the setup wizard with:
aws configure agent-toolkit
The CLI detects supported coding agents installed on your system and helps configure the toolkit.
After setup, you can restart your coding agent and start asking AWS-specific questions.
For example:
What AWS services should I use to build a serverless API?
Or something more practical:
Create an S3 bucket with versioning enabled and add a lifecycle policy.
The agent can find relevant AWS skills automatically instead of requiring you to manually search for the correct skill first.

Why is this different from simply asking an AI about AWS?

This is probably the most important part.
A general AI coding assistant can generate something that looks correct.
For cloud infrastructure, that is not always enough.
AWS services change frequently. New APIs appear, recommended patterns change and security configuration matters. An outdated suggestion can still produce perfectly valid-looking code.
The Agent Toolkit gives the coding agent access to AWS-specific, maintained guidance rather than making it depend entirely on general model knowledge.
That changes the role of the coding assistant.
Instead of:
Developer → AI model → guessed AWS implementation
the workflow moves closer to:
Developer
↓
Coding Agent
↓
AWS Skills + MCP
↓
AWS APIs and services

Where I think this becomes useful

Consider a developer building a serverless application.
They may need:
  • API Gateway
  • Lambda
  • DynamoDB
  • IAM
  • CloudWatch
  • Infrastructure as Code
The difficult part is often not writing one Lambda function. It is connecting all these services correctly while keeping permissions, configuration and deployment in mind.
An AWS-aware coding agent could help across that complete workflow instead of only generating application code.
This could be especially useful while learning AWS because you can ask questions in the context of the project you are actually building.

But there is still an important rule

Giving an AI agent access to AWS does not mean we should blindly approve everything it generates.
IAM permissions, infrastructure changes, production resources and anything that can affect cost still need developer review.
AI can reduce the amount of manual work required to understand and configure AWS, but responsibility for what gets deployed still belongs to the developer.
In fact, learning to review AI-generated cloud infrastructure may become just as important as learning to generate it.

Final thought

For me, the interesting part of the Agent Toolkit is not that AWS added another AI feature.
It is that the boundary between coding assistant and cloud development environment is starting to disappear.
If coding agents can understand current AWS services, access the right tools and follow AWS-specific guidance, we may spend less time telling AI how AWS works and more time describing what we actually want to build.
That is a much more useful direction for AI-assisted cloud development.

References

  • AWS Agent Toolkit for AWS documentation
  • AWS CLI Agent Toolkit documentation
  • AWS Developer Tools Blog: Get started with the Agent Toolkit for AWS in the AWS CLI
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