On November 11 Amazon launched AWS Emulator, a local emulator of the AWS APIs. The AWS CLI, the SDKs and coding agents point at 127.0.0.1:9297 through a few environment variables instead of the real cloud, with no AWS account and no extra fee. It is written in Rust and generated from the same API models the AWS SDKs use; AWS says it covers more than 400 API models across more than 200 services. Most unusually, Senior Principal Engineer Matthew Goldberg says thousands of autonomous agents wrote, tested and fixed all of the code, and his team neither wrote nor reviewed a single line. The source and the fidelity test suite are on GitHub under CC0.
Key Takeaways
- ✓Covers 200+ AWS services and 400+ API models (AWS figures) on a local HTTPS endpoint at 127.0.0.1:9297, with no AWS account and no fee
- ✓100% agent-written: three agent teams write tests, find divergences and fix bugs, with thousands of issues fixed per day by hundreds of concurrent agents (AWS figures)
- ✓The public fidelity suite manifest lists about 29,700 tests (our count), each verified against real AWS first; per-service k of n scores are published, with no overall pass rate given
- ✓Limits: no IAM policy evaluation, one account and one Region, verified only in us-west-2, in-memory state only; not for security testing or production
- ✓Install via RPM/Debian packages on Linux or Docker on macOS/Windows, or run aws agent-toolkit add-skill --skill-name aws-emulator so a coding agent sets it up (needs AWS CLI 2.35.0+)
Key Decision Metrics at a Glance
Heavy Claude Code use: compare subscription limits and API bills
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and the pain point
When a coding agent builds an AWS app, every check means creating real cloud resources: slow, billable and risky. Third-party emulators such as LocalStack have filled the gap with uneven coverage and fidelity. AWS now ships its own local copy for developers and agents, backed by same-day posts from Werner Vogels and Swami Sivasubramanian.
How it works
Goldberg writes that it began as a weekend experiment. Inspired by Anthropic's agent-built C compiler, he had Kiro write tests from the AWS CLI, make them pass against real AWS, then build an in-memory implementation that passed the same tests; a usable SQS appeared in 15 minutes. The product is built by three agent teams on Bedrock AgentCore: one generates fidelity tests from the public API models, one replays them and files issues, and one fixes each issue. Every change must pass a merge gate covering coding standards, adversarial security review and no fidelity regressions.
Benchmarks
AWS measures fidelity as k of n tests passed per service, published with each GitHub release. By our count the evaluation suite manifest holds about 29,700 tests. AWS gives no overall pass rate and there is no independent comparison with LocalStack yet. AWS says control plane APIs match most closely, while data plane and CloudFormation, Terraform and CDK support vary by service. An IDC analyst notes it does not yet simulate throttling, outages or Regional differences.
Getting started
On Linux install the RPM or Debian package and run aws-emulator; on macOS and Windows build an image from the official Dockerfile. Point AWS_ENDPOINT_URL and related variables at your machine and unset real credentials. Rust projects can use embedded mode with one isolated instance per test. Telemetry is on by default; turn it off with --telemetry off. Still test permissions and security on real AWS before production.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.