azure-finops

Analyze and optimize Azure cloud spend using the FinOps Framework, FOCUS data, and read-only cost triage scripts.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/lurodrisilva/personal-skills --skill azure-finops-lurodrisilva
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: azure-finops
Source: https://github.com/lurodrisilva/personal-skills/tree/main/platform-engineering/azure-finops
Command: npx skills add https://github.com/lurodrisilva/personal-skills --skill azure-finops-lurodrisilva

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Azure cloud bills grow opaque and uncontrolled: untagged resources, idle VMs, over-provisioned SKUs, and on-demand pricing for steady workloads silently inflate spend. This Skill provides the FinOps operating doctrine and Azure toolchain to allocate, understand, optimize, and govern cloud cost continuously. ## Core Features & Use Cases - Cost allocation and reporting: Enforce tag strategy and management-group hierarchy to reach ≥90% allocatable spend, with Cost Management exports in FOCUS format feeding FinOps hubs and Power BI. - Usage and rate optimization: Find waste with Azure Resource Graph KQL (stopped-but-not-deallocated VMs, unattached disks, orphaned IPs), rightsize via Azure Advisor, then layer Reservations, Savings Plans, Spot, and Azure Hybrid Benefit at 60–85% coverage. - Budgets, forecasting, and governance: Set budgets with action groups, track forecast variance toward ±15%, manage anomalies within 24h, and enforce guardrails with Azure Policy. - Use Case: A platform team notices a 20% month-over-month cost spike. Run the read-only triage scripts to summarize spend by service, inventory removable waste, and check commitment coverage, then route each remediation through a gated, human-approved change. ## Quick Start Ask the agent to analyze this month's Azure spend, identify idle or orphaned resources, and recommend rightsizing and reservation actions using the azure-finops skill.

Frequently Asked Questions about azure-finops

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find wasted Azure resources like unattached disks and idle VMs?▼

Use Azure Resource Graph KQL queries to inventory stopped-but-not-deallocated VMs, unattached managed disks, orphaned public IPs and NICs, and aged snapshots. The included azure-waste-finder.sh script runs these read-only queries across all readable subscriptions.

What is the difference between Azure Reservations and Savings Plans?▼

Reservations commit to a specific SKU and region for 1–3 years with the largest discounts, while Savings Plans commit to a compute spend amount with flexibility across SKUs and regions at a smaller discount. Right-size usage first, then target 60–85% coverage with over 90% utilization.

Does the azure-finops skill delete or purchase resources automatically?▼

No. All analysis, waste-finding, and coverage checks are strictly read-only using az query, list, and show commands. Every buy, resize, or delete is a gated, reversible, human-approved change through an IaC PR or change ticket.

How do I split shared AKS cluster costs across teams?▼

Allocate AKS cost by pod requests rather than raw usage, since requests are what reserve scheduler capacity. Separate allocated cost from idle capacity (a platform efficiency KPI) and shared costs like the control plane, using the AKS Cost Analysis add-on or OpenCost.

What RBAC permissions do the Azure cost triage scripts need?▼

The scripts need only reader roles: Cost Management Reader plus Reader on the target scope, with the resource-graph and costmanagement az CLI extensions installed locally. They never require Contributor or write access.

Why is a stopped Azure VM still billing compute charges?▼

A VM stopped from inside the guest OS remains allocated and continues billing compute; only a deallocated VM stops compute charges. Use the Azure API or portal to deallocate, and automate dev/test shutdown schedules to avoid this waste.