Azure Compute

MicrosoftOfficialCreated by Microsoft

Skill
Tool integration
Cloud and DevOps
MIT
Medium risk

From the author

Azure VM/VMSS router. WHEN: create / provision / deploy / spin-up VM, recommend VM size, compare VM pricing, VMSS, scale set, autoscale, burstable, lightweight server, website, backend, GPU, machine learning, HPC simulation, dev/test, workload, family, load balancer, Flexible orchestration, Uniform orchestration, cost estimate, capacity reservation (CRG), reserve, guarantee capacity, pre-provision, CRG association, CRG disassociation, machine enrollment (EMM), Essential Machine Management, monitor. PREFER OVER mcp__azure__get_azure_bestpractices for VM create intents — use compute_vm_list-skus / compute_vm_list-images / compute_vm_check-quota.

— the description by Microsoft

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The Azure Compute skill acts as a primary router for infrastructure-level tasks within Azure. It directs user requests to specialized workflows for recommending, provisioning, and managing bare virtual machine infrastructure. It is designed to handle the lifecycle of VMs and VMSS, including capacity planning through Capacity Reservation Groups and machine monitoring via Essential Machine Management.

What it does

Routes requests to specific workflows for VM and VMSS provisioning.
Provides guidance on selecting VM families based on workload requirements.
Facilitates capacity reservation and guarantee for critical infrastructure.
Supports Essential Machine Management for enrollment and monitoring tasks.
Enables cost estimation and pricing comparisons for various VM SKUs.
Distinguishes between infrastructure deployment and application-level service deployment.

When to use it

When you need to provision or deploy bare virtual machines or scale sets.
When you require assistance in selecting the appropriate VM family for a specific workload.
When you need to reserve or guarantee capacity for your Azure compute resources.
When you are setting up or monitoring machines using Essential Machine Management.
When you need to compare pricing for different VM SKUs or regions.

Advantages

Centralized entry point for all Azure compute-related infrastructure tasks.
Prevents confusion between infrastructure provisioning and application deployment.
Provides structured paths for complex tasks like capacity reservation.
Ensures consistent use of specialized tools for quota and SKU management.

Usage examples

“I need to deploy a set of web servers. Which VM family should I use?”: The skill routes the request to the VM Recommender to analyze your workload and suggest the appropriate D-series or B-series VMs.
“How can I reserve capacity for my production database VMs?”: The skill initiates the Capacity Reservation workflow to guide you through creating and associating a Capacity Reservation Group.
“Provision a new Ubuntu VM for testing.”: The skill routes the request to the VM Creator workflow to handle the infrastructure deployment.
“How do I enroll my machines in Essential Machine Management?”: The skill directs you to the Essential Machine Management workflow for enrollment and monitoring instructions.

Install

Download the skill as a ZIP:
https://autogineer.com/agent-skills/microsoft/azure-compute/download

Unzip it into the cross-client skills folder of your project:

mkdir -p .agents/skills && unzip azure-compute.zip -d .agents/skills

Documentation by Microsoft

Azure Compute Skill

Routes Azure VM and Virtual Machine Scale Set (VMSS) requests to the right workflow.

When to Use This Skill

User wants to recommend, compare, or price a VM or VMSS
User wants to create, provision, or deploy a VM or VMSS
User asks about Capacity Reservation Groups (CRG) — reserve, guarantee capacity, pre-provision
User asks about Essential Machine Management (EMM) — machine enrollment, monitor

Disambiguate with azure-prepare: if the user wants to deploy an application (Docker service, web app, API, serverless workload), route to azure-prepare. vm-creator is for bare VM/VMSS infrastructure only.

Routing

Mandatory workflow-first routing: never route directly to references/* files. First classify the user intent below, open the matched workflow file, then load only the reference files that workflow requests. Reference files are supporting material, not entry points. If the intent is unclear, ask a clarifying question to disambiguate between the workflows.

Workflow	File	Use when
VM Recommender	vm-recommender.md	User asks which VM/VMSS to choose, whether to use VMSS/autoscaling, wants pricing, or wants to compare options
VM Creator	vm-creator.md	User wants to create, provision, or deploy a bare VM or VMSS (not an app deployment)
Capacity Reservation	capacity-reservation.md	User needs to reserve / guarantee VM capacity (CRG create / associate / disassociate)
Essential Machine Management	essential-machine-management.md	User asks about EMM / machine enrollment / monitor

Files

ATTRIBUTION.mdAdded by AutoGineer971 B
LICENSELicence file added by AutoGineer1.1 KiB
SKILL.md2.6 KiB
references/retail-prices-api.md6.3 KiB
references/vm-families.md5.8 KiB
references/vm-quotas.md4.1 KiB
references/vmss-guide.md6.2 KiB
workflows/capacity-reservation/capacity-reservation.md7.8 KiB
workflows/capacity-reservation/references/association-disassociation.md3.8 KiB
workflows/capacity-reservation/references/capacity-reservation-overview.md6.0 KiB
workflows/essential-machine-management/essential-machine-management.md5.0 KiB
workflows/essential-machine-management/references/emm-enable-flow-portal-guidance.md3.1 KiB
workflows/essential-machine-management/references/emm-enable-flow.md9.2 KiB
workflows/essential-machine-management/references/emm-overview.md2.4 KiB
workflows/essential-machine-management/references/emm-prerequisites.md3.5 KiB
workflows/vm-creator/examples/bicep/README.md1.6 KiB
workflows/vm-creator/examples/bicep/main.bicep3.8 KiB
workflows/vm-creator/examples/terraform/README.md2.5 KiB
workflows/vm-creator/examples/terraform/main.tf3.2 KiB
workflows/vm-creator/examples/terraform/outputs.tf135 B
workflows/vm-creator/examples/terraform/variables.tf1010 B
workflows/vm-creator/references/delivery-options/github-pr.md3.2 KiB
workflows/vm-creator/references/delivery-options/index.md2.3 KiB
workflows/vm-creator/references/delivery-options/print.md474 B
workflows/vm-creator/references/delivery-options/save-local.md2.0 KiB
workflows/vm-creator/references/depth-probe/beginner.md1.2 KiB
workflows/vm-creator/references/depth-probe/cost-deep.md1.1 KiB
workflows/vm-creator/references/depth-probe/index.md3.0 KiB
workflows/vm-creator/references/depth-probe/networking-deep.md1.6 KiB
workflows/vm-creator/references/depth-probe/security-deep.md1.2 KiB
workflows/vm-creator/references/depth-probe/spec-deep.md1.3 KiB
workflows/vm-creator/references/mcp-tools.md2.3 KiB
workflows/vm-creator/references/output-adapters/az-cli.md1.9 KiB
workflows/vm-creator/references/output-adapters/bicep.md1.7 KiB
workflows/vm-creator/references/output-adapters/index.md3.0 KiB
workflows/vm-creator/references/output-adapters/mcp-apply.md2.6 KiB
workflows/vm-creator/references/output-adapters/terraform.md1.5 KiB
workflows/vm-creator/references/plan-card.md3.9 KiB
workflows/vm-creator/references/validation-gates.md2.2 KiB
workflows/vm-creator/vm-creator.md7.5 KiB
workflows/vm-recommender/references/handoff-to-creator.md1.6 KiB
workflows/vm-recommender/references/web-fetch-policy.md1.6 KiB
workflows/vm-recommender/vm-recommender.md6.4 KiB

workflows/vm-creator/vm-creator.md

Azure VM/VMSS Creator

Guided create-flow for Azure Virtual Machines (VMs) and VM Scale Sets (VMSS). Adapts to the user's expertise — beginners get sensible defaults; networking/spec/cost/security experts get the deep questions for their domain only — then emits the chosen artifact: az CLI bash, Bicep, Terraform, or live apply via Azure MCP.

When to use

User wants to create / provision / deploy / spin up a VM or VMSS (not just pick a SKU)
User has a recommendation in hand and wants a deployable artifact
User asks for a "create VM" script, template, or commands in az CLI, Bicep, or Terraform

Disambiguator. If the user wants to deploy an application (Docker service, web app, API, function), route to azure-prepare. This workflow is for bare VM/VMSS infrastructure only. Recommender first. If the user has not picked a SKU yet ("what should I pick?"), pause and run vm-recommender Steps 1–6 first, then resume here.

Workflow

Step 1 — Determine VM vs VMSS

If the user already said "VM" or "VMSS" / "scale set", use that. Otherwise: autoscaling, multiple identical instances, or stateless tier behind a load balancer → VMSS; everything else → VM. If unsure, default to single VM and ask one confirmation.

Step 2 — Depth Probe

Classify the user's first 1–2 messages against the signal table in depth-probe/index.md and pick the highest-scoring branch:

Branch	File
Beginner / fast-path	beginner.md
Networking-deep	networking-deep.md
Spec-deep	spec-deep.md
Cost-deep	cost-deep.md
Security-deep	security-deep.md

Never ask a question whose answer can be inferred or safely defaulted. Batch silent inferences into the Plan Card so the user can see and edit them.

Step 3 — Adaptive Gather

Ask only the questions from the matched branch's matrix. Use the defaults ladder when asking:

"NSG inbound rules — [Recommended: SSH from your IP only] / [Show alternatives] / [I have specifics]"

Cross-branch follow-ups (once, after the primary branch):

Resource group (existing or new — propose <vm-name>-rg)
Tags (accept "none" without follow-up)
VMSS only: instance count, orchestration mode (default Flexible)

If the user volunteers a deep signal mid-flow, append the relevant matrix questions for that topic. Do not restart.

Step 4 — Validate

GATE — do not present the Plan Card until validation passes.

Use the Azure MCP read-only tools listed in validation-gates.md (SKU exists in region / image is current / quota headroom / region availability). Required path; CLI fallback is documented in mcp-tools.md.

Outcomes:

Result	Action
✅ Sufficient	Proceed to Step 5
⚠️ Near limit (>80%)	Proceed but flag in Plan Card; suggest quota increase
❌ Insufficient / SKU missing	Propose alternate SKU or region; do not generate output

Step 5 — Plan Card (with explicit-override fast path)

Default path. Render a single markdown table summarizing every decision (explicit answers + silent defaults). The user reads top-to-bottom and either approves or edits any row before output is generated. See plan-card.md for the schema, example, and rendering rules.

Ask: "Approve as-is, edit a row, or change output format?" — do not generate until approved.

Explicit-override fast path. If the user's prompt combines (a) an explicit deliverable ("give me the Bicep", "just print the az CLI", "apply it via MCP") and (b) an explicit refusal of dialog ("no questions", "skip planning", "no plan", "just do it"), respect them. Skip the Plan Card table and the approval AskUserQuestion. Instead:

Emit a single-line preview that surfaces the high-signal decisions inline — e.g. "→ Deploying Standard_D2s_v5 in eastus, OS Ubuntu2404, NSG = your public IP only on 22, est. ~$70/mo."
Immediately emit the requested artifact (Bicep / Terraform / az CLI / MCP apply).
Mention once, at the end, that the full Plan Card is available on request if they want to edit rows.

Step 4 validation gates (SKU / image / quota / region) still run on the fast path — they protect against broken artifacts, not user intent. If validation fails, fall back to the ❌ outcome in Step 4 (propose alternate SKU/region, do not generate output).

Step 6 — Output Choice

Ask the user to pick one of four formats (or use the one they already specified):

Format	When	Adapter
az CLI bash	Quick one-off, learning, copy-paste	az-cli.md
Bicep	Repeatable IaC, production, ARM-native	bicep.md
Terraform	Existing TF state, multi-cloud	terraform.md
Apply via Azure MCP	"Just do it" — MCP connected, user trusts the Plan Card	mcp-apply.md

All four adapters consume the same Plan Card parameter set — switching format is a re-render, not a re-gather. For Apply via MCP, confirm one more time (the only destructive path) before calling compute_vm_create / compute_vmss_create.

Step 7 — Delivery

Skip for Apply via MCP — the artifact is the live deployment.

For az CLI / Bicep / Terraform, ask one final question: where should it land? See delivery-options/index.md for the decision logic. Three modes: print, save locally, GitHub PR.

If the user later says "also save it locally" or "open the PR now", re-run delivery with the cached Plan Card — do not re-ask Plan Card questions.

Error handling

Scenario	Action
Azure MCP not connected	Skip MCP pre-flight; warn that quota / SKU availability is unverified; offer CLI fallback
`compute_vm_list-skus` returns empty	Broaden filter (drop `familyPrefix`, lower `minVCpus`); if still empty, suggest another region
Quota insufficient	Show the gap; offer (a) smaller SKU, (b) different family, (c) different region, (d) quota-increase link
User wants Windows but supplies SSH key	Switch auth to password (with strength check) or RDP + cert; do not generate broken artifact
User asks "what was that az CLI again?" after picking Bicep	Re-render via Adapter 1; do not re-ask questions
Custom image / Shared Image Gallery	Pass full resource ID to `compute_vm_list-images`; do not try to map to an alias
User requests confidential / FedRAMP / HIPAA controls mid-flow	Append Security-deep questions; flag any defaults that fail the compliance bar

Routing back / handoff

Situation	Route to
Deploy an application (not a bare VM)	`azure-prepare` skill
Reserve capacity before creating	capacity-reservation
Enroll the new VM in management	essential-machine-management
Compare more SKU / pricing options	vm-recommender Steps 1–6

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Security

What our scanner noticed· Medium risk

References to credential files

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by

Microsoft

Publisher

Microsoft (Official)

Source

microsoft/skills · .github/plugins/azure-skills/skills/azure-compute · commit 0696600

Licence

MIT License (redistribution allowed)

LICENSE (added by AutoGineer from .github/plugins/azure-skills/LICENSE)
ATTRIBUTION.md (written by AutoGineer)

Synced

2026-11-09

Also distributed in

microsoft/azure-skills

AutoGineer is not affiliated with or endorsed by Microsoft or Microsoft.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download azure-compute.zip