AWS Emulator: test AWS code locally without an account
Added NewCopyable prompt below
Published Last verified Released
Our verdict
The most useful thing here is a fast inner loop for agents and CI: a coding agent can create a DynamoDB table or S3 bucket, run its code against it, read the real AWS-style error and fix it before anything is deployed. AWS made it generally available on 11 Nov 2026; it is generated from the same API models as the AWS SDKs, returns 501 Not Implemented for operations it does not support yet, keeps state in memory only, and publishes per-service fidelity scores with each release. An independent hands-on by pensol on Zenn ran Terraform apply and destroy for S3, DynamoDB, SQS, SSM Parameter Store and IAM in GitHub Actions from the .deb package without Docker, reporting about 0.25 s startup and about 170 MB of memory. Limits: IAM policies are not enforced the way AWS does (the same write-up warns not to use it for permission tests, and AWS says not to use it to test security); many operations are still stubs (a code-level comparison by gologolo counted 8,362 stubbed operations across 436 service crates); telemetry is on by default; macOS and Windows need a Docker image you build from a short Dockerfile. LocalStack's own response (13 Nov) positions LocalStack as a wider platform with persistence, IAM policy automation, chaos testing and a web UI, which is where it still leads. We have not run it ourselves.
- Who it's for
- Developers and teams building on AWS who want fast, free integration tests in CI or a safe sandbox for coding agents, and LocalStack users reviewing their options
- Leverage
- AWS integration tests and agent fix-and-retry loops that run in seconds on a laptop or CI runner, with no AWS account, no cloud cost and no risk of touching real resources
- Effort
- Low
- Date added
What it does
AWS Emulator (aws/aws-emulator) is a Rust program with no GUI that serves an AWS API endpoint on https://127.0.0.1:9297 with a self-signed certificate. Each run generates its own access keys, an SDK config file and a CA certificate under ~/.aws/emulator/<port>/; requests must be SigV4-signed by default (turn off with --auth off). Each copy gets its own port, keys and in-memory state, so parallel test suites, CI jobs and agents do not see each other's resources. It runs as a Linux RPM or .deb package, in a Docker container on macOS and Windows, or embedded in Rust tests (one in-process instance per test, with a large first build). The aws-emulator agent skill installs through the AWS CLI (2.35.0 or later) and can check which AWS operations your repo uses against the published fidelity tests, then tell you which parts you can rely on and draft an issue for any difference, showing it to you before posting. The code is CC0-1.0 apart from vendor/ (Apache-2.0), and the repository does not take pull requests.
How builders use it
- Install the agent skill with the AWS CLI and ask your agent: "Will my app in this repo work on AWS Emulator?" before you rely on it.
- Start the emulator (aws-emulator on Linux, or the Docker image elsewhere) and set the environment variables from the docs so your CLI, SDK or Terraform points at it.
- Run your integration tests or terraform apply and destroy against it in CI on every pull request, and keep permission and security checks on a separate step (policy scanning or real AWS).
- Turn telemetry off with --telemetry off or EMULATOR_TELEMETRY=off if your team requires it.
What to expect
Infrastructure and API mistakes surface in seconds on every change instead of after a slow deploy, and agents can iterate on AWS code without credentials to real accounts.
Copy this: Agent skill + point your SDK at it
# Official source: github.com/aws/aws-emulator (CC0-1.0). Docs: docs.aws.amazon.com/emulator
# 1) Agent route (AWS CLI 2.35.0+):
aws agent-toolkit add-skill --skill-name aws-emulator --region us-east-1
# then ask your agent:
Will my app in this repo work on AWS Emulator? List the AWS operations it calls, which ones have passing fidelity tests, and which ones I should still test against real AWS.
# 2) Manual route (Linux package or Docker image), then in your shell:
aws-emulator --telemetry off
export AWS_ENDPOINT_URL=https://127.0.0.1:9297
export AWS_SHARED_CREDENTIALS_FILE=~/.aws/emulator/9297/emulator_credentials
export AWS_CONFIG_FILE=~/.aws/emulator/9297/config
export AWS_CA_BUNDLE=~/.aws/emulator/9297/ca.pem
export AWS_EC2_METADATA_DISABLED=true
export AWS_DISABLE_HOST_PREFIX_INJECTION=true
export AWS_REGION=us-east-1 AWS_DEFAULT_REGION=us-east-1
# Not for permission or security tests: IAM policies are not enforced like AWS.Related discoveries
See the newest picks →- ToolResearch & Automation
agent-desktop: computer use for agents via the accessibility tree
An Apache-2.0 Rust CLI and C library from Vercel Labs that gives any agent structured, retry-safe control of desktop apps through the operating system's accessibility tree.
Effort: LowPrompt#Open source#CLI
- ToolCoding & Shipping Agents
Sphinx: quiz yourself before shipping agent code
MIT Git hook that quizzes you on what your coding agent's change does before a protected push continues.
Effort: MediumPrompt#Plugin / extension#Open source
- ToolCoding & Shipping Agents
Shannon: open-source A.I. pentester for web apps and APIs
Keygraph's AGPL-3.0 A.I. pentester: it reads your web app's source, then attacks a running copy and only reports vulnerabilities it could actually exploit.
Effort: MediumPrompt#Open source#Self-hostable
- ToolCoding & Shipping Agents
Chrome DevTools MCP: let your coding agent debug in real Chrome
The Chrome DevTools team's official MCP server and CLI (Apache-2.0). It lets Claude Code, Cursor, Codex, Copilot, Gemini CLI or Antigravity open and control a live Chrome, read console messages with source-mapped stack traces, inspect network requests, take screenshots and record performance traces.
NewEffort: LowPrompt#Plugin / extension#MCP
Get the best new A.I. picks in your inbox
The best new agent workflows and tools we’ve vetted, with a clear verdict on each. Short, occasional, useful.