Nov 18, 20266 min read

AWS Proton Alternatives After the October 2026 Shutdown

AWS Proton reached end of support on October 7, 2026. What happened to the stacks it deployed, the options AWS recommends in its place, and how to give developers self-service infrastructure again without rebuilding Proton's templates.

Illustration for AWS Proton Alternatives After the October 2026 Shutdown

AWS Proton was AWS's answer to platform engineering: a platform team wrote environment and service templates in CloudFormation or Terraform, and developers picked a template, filled in a few parameters and got a running service with its pipeline. AWS ended support on October 7, 2026, so the console is gone, Proton's records of templates and service instances have been deleted, and the CloudFormation stacks it created are still running with nothing left that knows how to update them as a group.

The problem to solve now is the one Proton was bought for: letting product teams create and change services without each one becoming an infrastructure project, while the platform team keeps control of networking, IAM and defaults. This guide covers what changed for existing stacks and compares the alternatives AWS recommends with an approach where the application declares its own infrastructure.

AWS Proton Alternatives at a Glance

EncoreCloudFormation Git syncHarmonix on AWSCodePipeline and CodeBuildGitHub Actions
Developer interfaceInfrastructure declared in TypeScript or GoParameter files in a forked repoBackstage portal and templatesPipelines the platform team buildsWorkflows in the repo
Templates to maintainNone per serviceCloudFormation templatesBackstage templates plus IaCCloudFormation, CDK or TerraformCloudFormation, CDK or Terraform
EnvironmentsBuilt in, with preview environments per pull requestNo concept of environmentsYesthrough the AWS pluginBuilt by youBuilt by you
Local developmentencore run starts databases, queues and servicesNoNoNoNo
Maintained byEncore, open source frameworkAWSAWS Partner reference implementationAWSGitHub
Best forTeams that want self-service without per-service templatesSmall CloudFormation shopsEnterprises wanting a developer portalTeams wanting full pipeline controlSmall teams on GitHub

What Happened to Your Proton Stacks

Proton was a delivery layer on top of CloudFormation, so every environment and service instance it deployed is still a CloudFormation stack in your account, with its VPC, ECS services, Lambda functions or pipelines intact. What disappeared is the mapping from template versions to those stacks and the pipelines Proton managed.

The first step is an inventory: list the CloudFormation stacks Proton created, decide which ones still matter, and move each into a tool that can update it. Stacks you plan to keep as they are can be managed directly in CloudFormation, while services you are about to change are often a good moment to move to whatever replaces Proton.

1. Encore

Encore

Encore brings infrastructure from code, which replaces the service-template half of Proton with something developers already use: the application declares the databases, topics, buckets and cron jobs it needs, and the Encore platform provisions them in your AWS account. Encore provisions an IAM role for the service. A platform team sets defaults once, at the environment level, instead of versioning a template per service type, as the infrastructure configuration docs describe.

Set up the app

Get started
brew install encoredev/tap/encore &&
encore app create
Copy

Run in your terminal to get started locally.

~
$ encore app create internal-catalog --example=ts/hello-world$ cd internal-catalog && mkdir -p catalog/migrations

A Proton service template for "API with a database and a queue" becomes a few declarations in the service. The catalog folder has an encore.service.ts exporting new Service("catalog") and a migration creating an items table with sku TEXT PRIMARY KEY, name and price_cents:

// catalog/catalog.ts import { api, APIError } from "encore.dev/api"; import { SQLDatabase } from "encore.dev/storage/sqldb"; import { Topic } from "encore.dev/pubsub"; // What the Proton template's RDS section used to create. const db = new SQLDatabase("catalog", { migrations: "./migrations" }); interface PriceChanged { sku: string; priceCents: number; } // What the template's SNS and SQS section used to create. export const priceChanged = new Topic<PriceChanged>("price-changed", { deliveryGuarantee: "at-least-once", }); export const setPrice = api( { expose: true, method: "PUT", path: "/items/:sku/price" }, async ({ sku, priceCents }: { sku: string; priceCents: number }): Promise<void> => { if (!Number.isInteger(priceCents) || priceCents < 0) { throw APIError.invalidArgument("priceCents must be a non-negative integer"); } const row = await db.queryRow<{ sku: string }>` UPDATE items SET price_cents = ${priceCents} WHERE sku = ${sku} RETURNING sku`; if (!row) throw APIError.notFound("unknown sku"); await priceChanged.publish({ sku, priceCents }); }, );

encore run starts Postgres and the topic locally, which Proton never offered, so a developer tests the service before any stack exists. On push, Encore provisions the service on ECS, the RDS database and an SNS topic in a VPC it creates, and Encore provisions an IAM role for the service. Each pull request can get its own preview environment. A coding agent asked to add a subscriber or a bucket changes the code and the infrastructure in the same pull request, inside the defaults the platform team set, without editing a template.

Encore manages new services it provisions, and existing Proton stacks stay in CloudFormation unless you rebuild them as Encore services. Encore can also import an existing RDS database into an app.

Best fit: teams that adopted Proton to give developers self-service infrastructure and want that without maintaining a library of service templates.

2. CloudFormation Git Sync

Git sync connects a CloudFormation stack to a template and parameter file in a Git repository, and updates the stack when that file changes. A platform team keeps the template in a repo that developers fork and parameterize, which is the closest experience to Proton.

AWS's own guide notes it has no concept of environments and no advanced pipeline features, and relies on GitHub-style workflows. Template versioning and rollout across many services stay manual.

Best fit: small teams already fluent in CloudFormation who want a simple GitOps flow.

3. Harmonix on AWS

Harmonix on AWS is an AWS Partner reference implementation built on Backstage, with an AWS plugin for templates, environments and services similar to Proton's model. It gives developers a portal with a software catalog and scaffolding.

It is a reference implementation rather than an AWS service, so you deploy, customize and maintain it, along with Backstage itself and the IaC templates behind each scaffolder action.

Best fit: enterprises that want a developer portal and have a team to run Backstage.

4. CodePipeline and CodeBuild

Building delivery on CodePipeline and CodeBuild directly gives the most control: each service gets a pipeline that runs CDK, CloudFormation or Terraform, with approvals and stages you design. It is actively developed and integrates with the rest of AWS.

The cost is implementation work and less self-service, since there is no catalog or template selection unless you build one.

Best fit: platform teams that want full control of pipelines and are prepared to build the developer interface.

5. GitHub Actions

For teams on GitHub, Actions workflows with OIDC federation to AWS can run CDK, CloudFormation or Terraform on each push. It is simple to start and has a large marketplace of actions.

Platform controls, such as which templates and parameters a team may use, need extra work through reusable workflows, branch protection and policy checks.

Best fit: smaller teams already on GitHub who want delivery close to the code.

How to Choose

Pick CloudFormation Git sync if you want the nearest thing to Proton with the least new tooling. Pick Harmonix if a developer portal is the goal and you can run Backstage. Pick CodePipeline or GitHub Actions when you would rather build exactly the pipeline you need. Pick Encore when the reason you adopted Proton was developer self-service, and you would rather have services declare their infrastructure in code than maintain templates for them.

Frequently asked questions

Is AWS Proton discontinued?

Yes. AWS ended support for AWS Proton on October 7, 2026, after closing it to new customers on October 7, 2025. The Proton console and Proton resources are no longer accessible, and Proton data was deleted after that date.

What happens to infrastructure deployed by AWS Proton?

The infrastructure keeps running. Proton deployed CloudFormation stacks, and those stacks and the resources they manage remain in your account. What is gone is the Proton layer: templates, environments, service instances and the pipelines that updated them, so changes now have to go through CloudFormation directly or a new delivery tool.

What does AWS recommend instead of Proton?

AWS's deprecation guide lists CloudFormation Git sync for a GitOps workflow, Harmonix on AWS for a Backstage-based developer portal, CodePipeline with CodeBuild for maximum flexibility, and GitHub Actions for smaller teams already on GitHub.

How can developers get self-service infrastructure without Proton templates?

One option is to let the application declare what it needs. With Encore, services declare their Postgres databases, Pub/Sub topics, buckets and cron jobs in TypeScript or Go, and a git push provisions them in your AWS account with the same networking, IAM and backup settings for every service, so platform rules live in the platform rather than in per-service templates.

Deploy to your cloud on AWS and Google Cloud
without managing infrastructure

Encore lets developers and agents define application infrastructure directly in code, then automatically provisions it from local development to production in your AWS or GCP account.

~/orders — claude
Claude Code
Claude Code v2.1.180Opus 5.5~/orders