Automate deployment.
The gap between "tests pass on my machine" and "it's running in production" is where deploys go to die. Most teams cross it with a hand-assembled pipeline someone wired up once, half-remembers, and nobody wants to touch. This blueprint writes that pipeline for you, as code in your repo, so the path from a green test suite to a running container on ECS is one file you can actually read.
A GitHub Actions workflow that runs the full path on every push. It executes your Jest test suite first, and only if tests pass does it build a Docker image, push that image to Amazon ECR, and trigger a deployment on ECS. The steps are ordered so a broken build never reaches production, and the whole workflow lives in .github/workflows where your team can review it.
You watch it happen in the task workspace: the agent scaffolds the workflow file, wires the ECR and ECS steps, and walks through the pipeline stage by stage before handing it over. The code lands on a branch in your repository with a pull request ready.
The first version covers the core pipeline and is yours to steer with plain follow-up messages:
Each follow-up wakes the same repository and the agent continues on its own code.
Do I need to know how to code? No. You describe the pipeline you want; the agent writes the workflow, and the pull request shows you exactly what runs and when.
Where do my AWS credentials go? Into your repository's encrypted GitHub secrets, referenced by name in the workflow. They're never written into the code the agent commits.
Can I test it before it touches production? Yes. Ask for the ECS deploy to be gated behind a manual approval or a protected branch, and nothing ships without your say-so.