AWS Emulator Runs 200+ AWS Services Locally, Coded by AI Agents
AWS Emulator is free, runs on your laptop and covers 200+ services, yet no human wrote or reviewed its code. Here is what it can test and where it falls short.
Hakkı ALKANPublished 5 min read
Amazon Web Services released AWS Emulator on Nov 11, 2026, a free local emulator that lets developers and coding agents test code against AWS APIs on their own machine instead of the cloud. According to the AWS News Blog, it has no additional service fee, is generally available now, and you pay only for AWS resources when you actually deploy.
The unusual part is who wrote it. In a Nov 11 post on About Amazon, Sr. Principal Engineer Matthew Goldberg said, "The emulator was written entirely by autonomous agents. Our team hasn't written or reviewed any of the code." The code is public on GitHub under a CC0 public domain dedication, with its full commit history.
What Is AWS Emulator and How Do You Run It?
AWS Emulator is a local stand-in for AWS. It is written in Rust to stay light on CPU and memory, and it serves HTTPS on 127.0.0.1:9297, so by default only your own computer can reach it. On Linux you install it from RPM or Debian packages; on macOS and Windows you build a Docker image from a short Dockerfile in the docs, because AWS does not publish a ready-made image. You can also build it from source.
The AWS CLI, the AWS SDKs and coding agents such as Kiro and Claude Code talk to it through environment variables like AWS_ENDPOINT_URL, so application code stays unchanged. A test that creates a DynamoDB table, writes an item and reads it back runs on the laptop without a deployment. For Rust projects, an embedded mode (the aws-emulator-embedded crate) runs the emulator inside the test process, and each instance keeps its own state, so tests can run in parallel.
How Close Is AWS Emulator to Real AWS?
The emulator is generated from the AWS API models, the same models the AWS SDKs use, so requests and error messages have the same shape as on AWS. GeekWire reports it is built from more than 400 API models covering more than 200 AWS services and works without a network connection. Goldberg says the project started as an empty web server that accepted requests for 438 API models.
Coverage is not equal across services. With each release, AWS publishes a fidelity score per service as "k of n", meaning how many of that service's fidelity tests (scripted workflows checked against real AWS) pass. An operation that is not implemented yet returns HTTP 501, so tests fail loudly instead of silently.
What Are the Limits at Launch?
- No real IAM: AWS says the emulator does not evaluate IAM (Identity and Access Management) and resource-based policies the way AWS does, so a request AWS would deny may succeed locally. AWS explicitly says not to use it to test whether an app is secure.
- One account, no Regions: each emulator acts as a single account (000000000000), with no cross-account or cross-Region behavior.
- No failure simulation yet: IDC analyst Devin Pratt told InfoWorld, as cited by GeekWire, that it does not yet simulate throttling, outages or regional service differences, and does not sandbox the code it runs.
- Telemetry on by default: it sends anonymous usage and error data unless you turn it off.
How Did AI Agents Build It?
Goldberg says it began as a weekend experiment: he gave Amazon's Kiro coding agent a simple prompt and got a passable Amazon SQS emulator in about 15 minutes. The product version is run by three teams of AI agents on Amazon Bedrock AgentCore. One writes fidelity tests from the public API models and verifies them against real AWS, one replays those tests against the emulator and files issues, and one fixes a single issue per agent.
Every change passes a merge gate with compile checks, coding standards reviews, adversarial security reviews and fidelity regression checks. Goldberg says thousands of issues are fixed every day as hundreds of agents work on the codebase at once, and the backlog of known gaps is public in a GitHub folder called "things-that-do-not-work".
| Question | AWS Emulator | Real AWS |
|---|---|---|
| Cost to test | No additional service fee | You pay for the resources you create |
| Where it runs | Your machine, 127.0.0.1:9297 | AWS Regions |
| IAM policy checks | Not evaluated like AWS | Authoritative |
| Accounts and Regions | Single account, no cross-Region | Multi-account, multi-Region |
What Does AWS Emulator Mean for Developers?
Our take: the timing is pointed. GeekWire notes that LocalStack, the best-known third-party AWS emulator, put its free version behind an account requirement this year and limited it to noncommercial use. A free, public domain emulator from AWS itself, with published fidelity scores, is a direct answer, and it fits a world where coding agents run hundreds of test loops an hour. The catch is that it is a fast feedback tool, not a final exam; security and failure behavior still need real AWS.
The agent angle also deserves skepticism. "No human reviewed the code" is only as safe as the tests and gates around it, which is why the public test suite matters more than the headline. Coding agents handling cloud credentials have already caused trouble, as our report on Claude Code secrets leaking to GitHub showed.
Before you switch your test suite over:
- Check the fidelity score for every AWS service you depend on.
- Unset real AWS credentials in test jobs; the emulator only accepts keys that start with "EMULATOR".
- Keep IAM, throttling and outage tests on real AWS.
- Turn off telemetry if your policy requires it.
Relative test: if your current tests spin up a DynamoDB table or SQS queue in a shared dev account for every run, AWS Emulator is worth trying today. If your bugs usually come from permissions or Region quirks, it will not catch them yet.