# AWS releases aws-emulator for local tests, written in Rust by a swarm of agents

> AWS has released an official emulator that lets developers run tests against a local copy of its services without a paid AWS API key.

- Status: official (From the company, authors or authority themselves, with a link to their own page.)
- Topic: AI in Practice
- Published: 2026-11-11T21:18:30+00:00
- Updated: 2026-11-11T22:09:10.977721+00:00
- Web page: https://ai-news.ikangai.com/story/aws-releases-aws-emulator-for-local-tests-written-in-rust-by-a-swarm-of-agents/
- Seen in: Simon Willison
- Change: capability (New capability)
- Applies to: aws-emulator

## What changed

AWS released aws-emulator, an official local emulator that lets developers run tests against a local copy of AWS services without provisioning a paid AWS API key.

AWS has released aws-emulator, an official emulator that lets developers run tests against a local copy of its many services without provisioning a paid AWS API key. It is a local program that serves an AWS API endpoint and has no graphical interface. The project's README says there is no additional service fee, so developers pay only for the AWS resources they deploy. It also says the emulator is meant for building and testing, not for production workloads or security testing.

The launch post also explains how the code was produced. Matthew Goldberg, a Sr. Principal Engineer on the AWS Developer Tools team, wrote that the emulator was built entirely by autonomous agents. The post says: "Our team hasn't written or reviewed any of the code." The GitHub repository shows 29,684 commits, and Simon Willison, who flagged the release, cites the same figure for the first three weeks of the project. Willison describes the code as written in Rust by a huge swarm of agents.

Goldberg's account starts with a weekend experiment. He wrote that AWS has two things that let agents build a faithful implementation of a system: a written spec, found in the public AWS API models, and a reference implementation, the real AWS cloud. He gave Kiro a short prompt asking it to write tests based on the AWS Command Line Interface, iterate until they pass against real AWS, and then build an in-memory implementation that passes the same tests. Kiro returned a first version of Amazon Simple Queue Service after 15 minutes, and over the weekend it produced more services. Goldberg says the result had gaps but was compelling enough that his team began using it for testing the Monday after.

Turning that prototype into a shipped product took more than a weekend. Goldberg says three groups of agents do the work. One team writes fidelity tests, each a real-world AWS scenario verified against real AWS before it is submitted, and the company has tens of thousands of them. A second team replays those tests against the emulator and files issues for failures, grouping failures that may share a root cause. A third team takes one issue at a time, finds the cause, and edits the code. Each change must pass a merge gate that checks compilation, coding standards, security, and whether fidelity regressed. Goldberg says thousands of issues are fixed each day by hundreds of agents working concurrently. The initial repository skeleton, he wrote, had a web server that accepted requests for 438 AWS API models but did nothing with them.

Werner Vogels' blog carries a guest post by Swaminathan Sivasubramanian that frames the same work around testing distributed systems. Sivasubramanian describes a distributed lock service called Alf, which was first tested in a simulated environment. Testing in a real environment then turned up problems the simulation could not reproduce. He argues that AWS now has more than 200 services, and that hand-building a test environment for each would take years. In his own experiment, he used Kiro to rebuild the replication engine he had built with a colleague for Alf, which had taken about four months the first time. He wrote that he supervised the agent for four nights before realizing it had no way to check its results. Once it had a spec and a test environment, the rebuild took about two hours. The lesson he draws is that the engineer's job shifts toward building specs, tests, and merge gates.

Some details of the project bear on the launch framing. The repository's issues folder lists behaviors that do not yet match AWS, and the Issues tab shows 14. The README says fidelity scores are published with each release. It also says the repository does not accept pull requests, and asks users to report bugs and fidelity gaps as GitHub issues. Goldberg says an earlier version had the agents build their own Python interpreter to support AWS Lambda functions, and that the team later gave them guidance to reuse existing libraries.

Installation depends on the platform. On Linux, users can install an RPM or Debian package. On Windows and macOS, the README says the emulator runs in a container built from a short Dockerfile, which is the Docker requirement Willison flagged. Building from source is supported only on Linux, on x86_64 and aarch64, and requires at least 100 GB of free disk space for the target directory. Telemetry is on by default and can be turned off with a command-line option, an environment variable, a configuration file, or the embedded API. Willison said he hopes AWS will eventually ship a single downloadable binary that other test suites on other platforms can use. The README says the code is released under the CC0 1.0 license, except the vendor directory, which is under Apache License 2.0.

## Why it matters

Teams that test AI and cloud workloads against AWS services could run those tests locally instead of paying for live API access. The agent-built codebase is also a concrete example of agentic engineering at scale.

## What you can do

Teams that run AWS-dependent test suites could evaluate the emulator to cut paid API usage in CI, noting that it currently needs Docker on macOS and Windows.

## Original source

- [AWS Emulator: a local emulator built 100% by autonomous AI agents](https://www.aboutamazon.com:443/news/aws/how-autonomous-ai-agents-built-aws-emulator)
- [GitHub - aws/aws-emulator: AWS Emulator is a local emulator of the AWS APIs. · GitHub](https://github.com/aws/aws-emulator)
- [AWS Emulator and the engineer's job when agents write the code | All Things Distributed](https://www.allthingsdistributed.com/2026/11/aws-emulator-and-the-engineers-job-when-agents-write-the-code.html)

## Further reporting

- [aws-emulator](https://simonwillison.net/2026/Nov/11/aws-emulator/)

---

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it.

Cite as: IKANGAI AI News, "AWS releases aws-emulator for local tests, written in Rust by a swarm of agents", 2026-11-11, https://ai-news.ikangai.com/story/aws-releases-aws-emulator-for-local-tests-written-in-rust-by-a-swarm-of-agents/
