AWS has made AWS Emulator generally available, giving developers a locally run environment for testing applications against supported AWS service APIs without deploying resources to an AWS account for each development cycle. The tool accepts requests from the AWS Command Line Interface (AWS CLI), AWS SDKs and coding agents through a local HTTPS endpoint.
AWS says the emulator has no additional service fee. Developers pay for AWS resources when they deploy on AWS. The project’s source code is available on GitHub. AWS supports installation through RPM or Debian packages on Linux; on macOS and Windows, users build and run it in Docker using a Dockerfile in the documentation. AWS does not publish a ready-made container image.
Local endpoint, familiar tools
After installation, a developer starts the emulator and configures command-line tools or application tests to use its endpoint instead of AWS. The emulator listens on 127.0.0.1:9297 by default, limiting access to the local machine. On startup, it generates credentials, a configuration file and a certificate under ~/.aws/emulator/9297/. Environment variables direct AWS tools to those files and the local endpoint, while application code can remain unchanged.
AWS describes a workflow in which the AWS CLI creates an S3 bucket, uploads an object and lists its contents. The same local setup can run a DynamoDB test that creates a table, writes an item and reads it back. A coding agent can also issue those operations and inspect their results locally, without waiting for a deployment in an AWS Region.
The emulator is implemented in Rust and is built from models in the AWS API models repository—the models also used by AWS SDKs. AWS says this approach gives requests and returned errors the same shape as their AWS counterparts. Supported APIs include those for services represented in the models, but implementation coverage varies by service and feature.
Two ways to run tests
For most tools, the emulator runs as a separate local process and serves requests over HTTPS. Requests must use AWS Signature Version 4 (SigV4). A health endpoint at /__admin/health allows scripts or containers to check readiness before starting tests. Some SDKs may need an additional configuration to trust the emulator’s certificate; AWS gives the Java SDK trust store as an example.
AWS also provides an embedded mode for tests written with the AWS SDK for Rust. Instead of contacting a separate server, a test creates an emulator instance in its own process and connects the SDK through an HTTP client. Each embedded instance maintains separate state, which AWS says allows tests to avoid sharing buckets, tables or queues and to run in parallel.
When the standalone process starts, its generated access key IDs begin with EMULATOR, and it rejects requests signed with other keys. AWS advises developers to remove real credential variables from test environments and configure the emulator’s generated credentials. The emulator uses a single account, identified as 000000000000.
Fidelity scores and limits
AWS measures compatibility through fidelity tests: SDK-driven workflows that are checked against actual AWS behavior. A test passes only when the emulator matches the response status, body, exact error code and message, as well as the resources and data created by the workflow. AWS publishes results for each service as a passed-test count out of the total, and makes the evaluation suite public on GitHub. The scores are snapshots; adding tests can lower a score even as implementation improves.
Unsupported operations return HTTP 501, or Not Implemented. AWS says developers can report behavior that differs from AWS through an issue in the emulator repository. Before relying on the tool for a particular service, teams should review its published fidelity results.
AWS explicitly cautions that the emulator is not a substitute for testing security on AWS. It does not evaluate IAM policies or resource-based policies as AWS does, so an operation denied in AWS might succeed locally. It also does not provide cross-account or cross-Region behavior. Developers should validate security settings and their complete application on AWS before production.
Agent-assisted development and maintenance
AWS says agents help maintain the emulator using AWS API models, Amazon Bedrock AgentCore Runtime Instances, AWS CodePipeline and AWS CodeCommit. One group identifies gaps by comparing responses with AWS and checking code against Smithy models; another works on fixes, tests and reviews. Changes are then sent through a CodePipeline run that executes fidelity tests before publication to the public GitHub repository. AWS also describes an independent security agent reviewing changes during development and as a release-pipeline check.
By making AWS API interactions available locally, AWS Emulator offers another option for build and test loops that would otherwise depend on deployments. Its practical value will depend on coverage for the services and behaviors a team uses, while AWS-hosted testing remains necessary for policy enforcement and production validation.