Take a stack that already works under `flisk up` to AWS with that host's own tooling - ECS/Fargate, EKS, App Runner, Lambda or EC2. It maps each service.json (ports, health checks, dependencies, env) to production artifacts written for the host, never the local `.autogenerated/` files. Use when the user wants to ship a working Flisk stack to AWS. Flisk itself stays local; for Helm, Argo CD or Kustomize pipelines use flisk-helm.
Installs into .claude/skills of the current project.
Are you the author of Flisk Deploy Aws?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/flisk-sh-flisk-deploy-aws)
---
name: flisk-deploy-aws
description: Take a stack that already works under `flisk up` to AWS with that host's own tooling - ECS/Fargate, EKS, App Runner, Lambda or EC2. It maps each service.json (ports, health checks, dependencies, env) to production artifacts written for the host, never the local `.autogenerated/` files. Use when the user wants to ship a working Flisk stack to AWS. Flisk itself stays local; for Helm, Argo CD or Kustomize pipelines use flisk-helm.
license: MIT
compatibility: Requires a working local Flisk stack first. AWS credentials and CLI recommended.
metadata:
author: flisk-sh
version: "1.0"
---
# Deploy to AWS
Flisk is local-only. This skill covers taking a validated local stack to AWS.
## Workflow
1. Confirm the stack works locally: `flisk doctor && flisk up <stack>`.
2. Extract the contract from `service.json` files (ports, dependencies, health checks, env).
3. Choose the AWS target:
- ECS/Fargate or EKS for container services
- App Runner for simpler web services
- Lambda + API Gateway for serverless
4. Generate or adapt production artifacts (Dockerfile, task definition, Helm chart, or CDK/Terraform) from the service.json — do not use the local `.autogenerated/` files.
5. Set secrets via AWS Secrets Manager or Parameter Store.
6. Deploy and verify health endpoints.
## Guardrails
- Never deploy the local Traefik/Tilt generated files.
- Keep production config separate from local service.json.
- Test the production image locally first if possible.