Writing

The First Official AWS Project 100% Written by Autonomous AI Agents: AWS Emulator Setup Guide

Amazon just released its official local AWS Emulator under a CC0 license. But the real revolution is that not a single line of code was written or reviewed by humans: 3 teams of autonomous AI agents built it. Here is how it works and how to set it up.

Furkan ÖzayNovember 12, 2026 · 2 min read
czCZAY.DEV // EDITORIALEN
AWS

#AWS · 2 MIN READ

FURKAN ÖZAYVOL. 2026 // #15E1

Amazon has officially launched AWS Emulator, an open-source local emulator for AWS APIs under a CC0 1.0 Universal license.

For years, developers working with AWS have wrestled with the fear of surprise cloud bills or the partial fidelity of third-party mocks like LocalStack. AWS Emulator allows developers to run AWS APIs locally—with zero cloud cost, without an AWS account, and without inputting credit card information.

However, the reason this announcement is historic is how it was created: As Amazon Sr. Principal Engineer Matthew Goldberg and Amazon CTO Werner Vogels revealed, not a single line of code was written or reviewed by human software engineers.

· · ·

1. Why Is the Version v1.0.322? The Autonomous Agent Factory

When developers notice the v1.0.322 tag on GitHub, many assume the tool has been public for months. In reality, Amazon unleashed three autonomous agent teams on Smithy (438 AWS API Models) over the past two months:

  1. Fidelity Test Team: Wrote tens of thousands of real-world integration tests against the actual AWS cloud.
  2. Divergence Detection Team: Replayed those tests against the local emulator, categorized failures by root cause, and filed automated GitHub issues.
  3. Bug-Fixing & Coding Team: Picked up issues individually, modified the code, and passed automated merge gates (compilation, coding standards, and security checks).

The agents iterated and produced 322 automated builds completely unassisted by humans before releasing v1.0.322 to the public.

· · ·

2. Installation & Quickstart

A) Add as a Skill to AI Coding Agents (Claude Code / Cursor)

With AWS CLI 2.35.0 or later installed:

Bash
aws agent-toolkit add-skill --skill-name aws-emulator --region us-east-1

B) Standalone Server Mode

Bash
# Start the local emulator on port 4566:
aws-emulator server --port 4566
 
# Route AWS CLI / SDK to the emulator:
export AWS_ENDPOINT_URL=http://localhost:4566
export AWS_ACCESS_KEY_ID=test
export AWS_SECRET_ACCESS_KEY=test
export AWS_DEFAULT_REGION=us-east-1
 
aws sqs create-queue --queue-name test-queue
aws s3 mb s3://test-bucket
· · ·

3. Important Boundaries

  • Do Not Test IAM / Permissions Boundaries: The emulator accepts functional parameters but does not strictly evaluate complex IAM boundary rules or cross-account policies.
  • For Building and Testing Only: Intended for local in-memory developer sandboxes, not persistent production workloads.

🔗 GitHub Repository: https://github.com/aws/aws-emulator

Read next