narvik889

Azure Acr Estimator — DevOps skill for Claude Code

DevOps community

Claude skill that pulls live Azure pricing via MCP and projects Annual Consumed Revenue (ACR) for a workload, faster than spending hours on the Azure calculator.

How to install Azure Acr Estimator

This entry records only its repository, not the path inside it, so there is no exact command to give. Open narvik889/azure-acr-estimator and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Azure Acr Estimator does

Claude skill that pulls live Azure pricing via MCP and projects Annual Consumed Revenue (ACR) for a workload, faster than spending hours on the Azure calculator.

Alternatives in DevOps

Datachain — The Context Layer for unstructured data: typed, versioned datasets over S3, GCS, Azure 2.8k ★
Azure Event Grid .net — Event Grid topic and domain publishing 1.8k ★
Chat — chat web app for teams, sass with user management and ratelimit, support chatgpt(openai & azure), claude, gemi 568 ★

README

Azure ACR Modeler — a Claude Skill

A [Claude Skill](https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview) that projects **Azure Consumed Revenue (ACR)** for a prospective workload using **live** Azure retail pricing — instead of an LLM guessing at numbers from memory, or a human spending twenty minutes to a few hours per estimate in the Azure Pricing Calculator.

Give it a rough workload description ("about 4 mid-size VMs and an App Service plan, East US"), and it resolves real prices, projects cost forward with a growth assumption, and outputs both a chat summary and a downloadable, auditable Excel workbook.

Read the full backstory — including the sandbox networking dead end and the App Service pricing gotcha that cost the most time — in the [blog post](#) that goes with this repo.

Why this exists

Estimating what a workload will cost on Azure is mechanical work: pull live pricing, apply a few reasonable assumptions, project it forward, hand it to whoever needs the number. Mechanical work done slightly differently every time is exactly where inconsistency creeps in — and an LLM prompted cold on the same task is non-deterministic by default. This skill fixes the assumptions and the workflow so the same input produces the same output every time, while still pulling real, current prices rather than stale training data.

What it covers today

Virtual Machines and Azure App Service compute costs
Pay-as-you-go (Consumption) retail pricing — not negotiated/EA pricing, reservations, or savings plans
12- or 36-month growth projections with a per-item monthly growth rate
An auditable Excel workbook (formulas, not pasted-in numbers) with Inputs / Projection / Summary tabs
Optional mapping to current Microsoft partner incentive tiers (looked up live, never hardcoded)

**Not yet covered:** storage, network egress, backup, monitoring, and other non-compute costs. See [`SKILL.md`](./SKILL.md) → "Known limits" for the full list.