AWS Released a Local Cloud Emulator — and AI Agents Wrote Every Line of Code

AWS shipped its AWS Emulator on November 11, 2026 — a free local testing tool that emulates more than 200 AWS services. The twist: not a single line of the codebase was written by a human. Autonomous AI agents wrote and reviewed every piece of it.

What Is the AWS Emulator?

AWS Emulator lets developers run their applications locally against faithful approximations of AWS APIs — S3, DynamoDB, Lambda, and hundreds of others — without an internet connection or a real AWS account. This cuts iteration time dramatically: deploy, test, fail, fix, all on your laptop.

At launch, it covers 438 AWS API models across more than 200 services, all published under a CC0 public-domain license (with vendor code under Apache 2.0).

Agents All the Way Down

AWS engineer Matthew Goldberg described the development model in stark terms: "The emulator was written entirely by autonomous agents. Our team hasn't written or reviewed any of the code."

The pipeline works like this:

  1. Test writers derive scenarios from public AWS Smithy API models and verify each against real AWS.
  2. Issue finders replay those tests against the emulator and file failures as issues.
  3. Fixers each take one issue and submit a patch.
  4. Security agents run adversarial reviews at the merge gate.

Humans set up the pipeline and defined the rules. After that, the agents iterated until the emulator passed its full test suite — entirely on their own.

Not for Production — But That's the Point

AWS is upfront about the limits. The emulator skips IAM permission checks and doesn't replicate cross-Region behavior, so you wouldn't use it to verify security posture. It's a local development tool: fast, offline, free, and disposable.

The recommendation from AWS is to run your full application against real AWS — including security settings — before any production deploy.

What This Signals

AWS Emulator is striking not just because of what it does, but because of how it was made. It's a production-grade, publicly shipped piece of infrastructure with zero human-written code. That's a milestone, and AWS chose to highlight the authorship rather than obscure it.

As agentic coding tools mature, AWS Emulator may be one of the clearest early examples of what "AI-native" software development actually looks like in practice: humans define constraints and quality bars, agents do the implementation, and the pipeline is the product.

The source code is available on GitHub under CC0, and you can start using it today for local development against any of the supported AWS services.