In Part 1 we set up a repository with spec-kit, a DevContainer, and azd, then built a /health endpoint from a spec, a plan, and a task list. This part starts where that one ended. Implementation is done, but two questions remain: what is still missing compared with the spec, and what do you do when someone reports a bug? We answer both in the same sample project.
Ever wondered how you can stop "vibe coding" and start building actual value from your ideas in a structured and repeatable way.
The fix isn't a better prompt — it's a better starting setup. Specifically: a repository structure that gives AI assistants structured context, pins the environment so every contributor runs identical tooling, and built in Azure deployment so there's no gap between "works locally" and "running in Azure."
This post shows you how by combining three tools into one workflow:
spec-kit — a CLI and VS Code workflow that turns vague feature ideas into structured specifications, implementation plans, and dependency-ordered task lists before a single line of code is written
DevContainers — a VS Code / GitHub Codespaces feature that pins the exact development environment in a .devcontainer/devcontainer.json committed to the repository
Azure Developer CLI (azd) — a CLI that standardises repository layout and enables one-command provisioning and deployment to Azure
Who this is for
Developers building on Azure who are comfortable with Git and VS Code but haven't yet established a structured, AI-assisted development workflow.
An active Azure subscription (required for the azd up step)
A GitHub Copilot license (Individual, Business, or Enterprise) if you want to use the cloud agent path in Step 8
By the end you will have a working repository: containerised for reproducibility, deployable to Azure Container Apps with a single command, and pre-wired for the full spec-kit SDD pipeline.
Growing up on a farm opened my eyes early on to how amazing and fragile nature is and how we as humans need to live in harmony with nature. Living in the Netherlands for the last couple of years has exposed me to awesome ways that farmers are pushing the boundaries to what is possible in "smart" technology driven agriculture. With that innovation comes the responsibility to keep ourselves accountable for our impact on nature. I believe that is where the Microsoft Planetary computer can play a role as real-time "planetary nervous system".
Whether you are just starting out in Azure, running many production workloads on a full Azure landing zone architecture, looking to migrate resources from on-prem to Azure or want to improve your hybrid cloud environment each of these phases comes with its set of challenges. In this article I will highlight some common pitfalls you can avoid in your cloud journey.
Series recap — Part 01: In Part 01 we covered the
security baseline for Azure Container Apps Environments: network isolation using a custom
VNet with dedicated subnets, locking down egress with Azure Firewall and UDR, disabling
public network access, and enabling internal-only ingress. We also walked through the
zero-trust identity model using Managed Identity and Azure Container Registry access.
In the 2nd part of this 2 part series on securing Azure Container Apps Environment, we will look into a deployment example. We will make use of Azure Verified Modules as far as possible. Then we will cover some scenarios on how you can make this available to your development teams and how development teams can perform deployments of Azure Container Apps into the managed environment safely.