Source: microsoft/azure-skills · Maintainer: Microsoft · License: MIT · Checked: 26 October 2026
What is the Azure prepare skill?
The Azure prepare skill makes a coding agent plan an Azure deployment before it writes any infrastructure, then generate the files the Azure Developer CLI needs: azure.yaml, Bicep or Terraform under ./infra/, and Dockerfiles. It is for developers who want an agent to take an app from a local folder to deploy-ready on Azure without guessing at services, regions or auth.
Its description begins: "Prepare azd-based Azure projects for deployment: generates azure.yaml, infrastructure (Bicep/Terraform), and Dockerfiles for the Azure Developer CLI (azd) workflow." It then says to use it only when you want azd as the deployment tool or the project already has an azure.yaml, and points Python App Service code-only deploys and cross-cloud migrations to other skills.
It ships in microsoft/azure-skills, the Azure Skills Plugin. The README lists it under "Build, deploy, and evolve" next to azure-validate and azure-deploy, and says the repo is synced automatically from microsoft/GitHub-Copilot-for-Azure. The SKILL.md frontmatter lists version 1.3.6 and author Microsoft.
As of 26 October 2026, its skills.sh page listed about 567,900 installs (snapshot, 26 October 2026).
What does it do?
The SKILL.md is strict. It opens with a banner headed "AUTHORITATIVE GUIDANCE" and "MANDATORY COMPLIANCE," and sets eleven rules, a plan-first workflow and a fixed hand-off.
Plan first. The agent's very first action must be writing a skeleton .azure/deployment-plan.md in the workspace root, not in a session folder. It fills the plan in as it analyzes the project, then shows it to you for approval. The skill says azure-validate and azure-deploy "depend on it and will fail without it."
Phase 1, planning:
| Step | What happens |
|---|---|
| 0 | Check for a specialized skill first (for example Python on App Service, AWS Lambda migration, AI gateway) |
| 1 | Pick a mode: NEW, MODIFY or MODERNIZE |
| 2 | Gather requirements: classification, scale, budget |
| 3 | Scan the codebase for components and dependencies |
| 4 | Select a recipe |
| 5 | Map components to Azure services |
| 6 to 7 | Finalize the plan file and ask you to approve it |
Phase 2, execution, only after approval: research each service, confirm subscription and region, generate artifacts, apply security hardening, run a functional check of the app (locally if possible), set the plan status to Ready for Validation, then invoke azure-validate.
Recipes. The recipe guide says "Default: AZD" and compares five options:
| Recipe | When the guide picks it | Deploy command |
|---|---|---|
| AZD (Bicep) | New and multi-service apps | azd up |
| AZD (Terraform) | Default whenever Terraform is mentioned | azd up |
| AZCLI | Existing az scripts, imperative control, AKS |
az commands |
| Bicep | IaC-first, no CLI wrapper | az deployment |
| Terraform | Multi-cloud or workflows incompatible with azd | terraform apply |
.NET Aspire projects are detected first and always use azd with azd init --from-code.
Service references. The folder carries guides for Container Apps, App Service, Functions (including Durable Functions and the Durable Task Scheduler), AKS, Static Web Apps, Cosmos DB, Azure SQL, Storage, Key Vault, Service Bus, Event Grid, Logic Apps, App Insights, API Management and Foundry, plus SDK notes for azd, Azure Identity and App Configuration.
Guardrails worth knowing:
- Subscription and region are always confirmed with you, never assumed.
- Destructive actions (delete, overwrite, cost impact, RBAC changes) need explicit confirmation.
- It never deletes your project folder, and
azd init -t <template>is for new, empty directories only. - SQL Server Bicep must never include
administratorLoginoradministratorLoginPassword; it uses Entra-only authentication.
How do you install the Azure prepare skill?
The repo README documents these routes. Install the whole plugin if you can, since this skill hands off to azure-validate and azure-deploy.
Claude Code: the Azure Skills Plugin. Run inside Claude Code:
/plugin install azure@claude-plugins-official
Or run /plugin and search for "azure". The plugin bundles all the Azure skills with the Azure MCP Server, which the README says gives agents "200+ structured tools across 40+ Azure services." Update with /plugin update azure@claude-plugins-official. See Claude Code plugins and Claude Code MCP.
GitHub Copilot CLI:
/plugin marketplace add microsoft/azure-skills
/plugin install azure@azure-skills
Codex and Cursor. The README says to run codex plugin marketplace add microsoft/azure-skills, then install azure from /plugins; for Cursor, use the Cursor Marketplace or Settings, then Plugins. See Codex skills and Cursor skills.
APM. The README also documents apm install microsoft/azure-skills for installing across several agents from one apm.yml.
Skill only. The command on skills.sh:
npx skills add https://github.com/microsoft/azure-skills --skill azure-prepare
This copies only this skill. Add azure-validate and azure-deploy the same way, or the hand-off at the end has nowhere to go.
Prerequisites, per the README: an Azure subscription, Node.js 18 or later for npx, the Azure CLI signed in with az login, and for deployment workflows the Azure Developer CLI signed in with azd auth login.
How do you use it?
Open the project folder and describe what you want built or hosted. The skill is model-invoked, so the agent can load it when you ask to prepare, host or create something on Azure with azd.
Example prompts (suggestions; the first is from the README's prompt list):
Prepare this app for Azure.Set up azd for this Node API and React front end. Use Container Apps and Terraform.Add a Service Bus queue and a queue-triggered function to this existing azd project.Create a timer-triggered Azure Function in Python that cleans old blobs every night, ready for azd.Modernize this Express app for Azure with azd. Use managed identity for the database.
Expect a plan file first and a request for approval. Read it closely: it is where the services, region and recipe get fixed.
Tips and limits
From the repo and skills.sh, not from testing:
- It writes files in your repo. Expect
.azure/deployment-plan.md,azure.yaml,./infra/and Dockerfiles. In MODIFY mode it edits existing files, so commit first and review the diff. - It is azd-first. The description scopes it to azd, and other skills take Python App Service deploys and AWS or GCP migrations.
- It does not deploy. After preparation it must run azure-validate; deployment belongs to azure-deploy.
- The plugin adds telemetry hooks. The Azure plugin's Claude Code hooks run a telemetry script on session start and after tool use. The README says to set
AZURE_MCP_COLLECT_TELEMETRY=falseto opt out. - Security audits. On skills.sh, Gen Agent Trust Hub and Snyk rated it Pass on 10 October 2026. Socket rated it Warn on 10 October 2026 with four alerts, one medium and three low. All four concern template code and docs, not malware: sample Cosmos DB endpoints with no authentication in the App Service recipes, and a Functions template guide that downloads remote templates and deploys them with
azd up. Socket's advice is to add authentication and input limits and to inspect templates before deploying.
The MIT License lets you copy and change the skill as long as you keep Microsoft's copyright and license notices.
Save it in your Kipt library
If you want your own edited copy, for example with your team's default region, naming rules or approved services written in, Kipt can hold it. Open Skills and paste the skill's GitHub folder link into the import box, or drop the folder or a .zip. A copy carries the skill files only, not the plugin's MCP configuration or hooks. Skills in your private GitHub repo install into Claude Code or Codex with one command:
npx skills add https://github.com/<you>/<repo>.git --all -g
Kipt is not affiliated with Microsoft.
Similar skills
- Azure validate skill: the next step, preflight checks on the files this skill generates.
- Azure deploy skill: runs the deployment once validation passes.
- Azure AI skill: Microsoft's skill for AI Search, Speech and Document Intelligence.
- Terraform skill: HashiCorp's style guide for HCL.
- Dockerfile skill: writing container images.
- Azure diagnostics skill: troubleshooting once the app is live.