Version 7.4.0 | Version info | Changelog
-0078D4?style=for-the-badge&logo=microsoftazure){kind=icon}
🔗 Shortlink: aka.ms/agenticinfraops
Azure infrastructure engineered by agents. Verified. Well-Architected. Deployable.
Agentic InfraOps is an IT Pro–focused workflow for building and operating Azure environments with guardrailed AI agents. It combines Microsoft’s *Azure MCP Server for live, RBAC-aware Azure context with a structured 7-step workflow, WAF-aligned decisioning, and AVM-first Bicep generation. The result: requirements → validated architecture → deploy-ready templates → as-built documentation and health checks, with fewer portal clicks and more repeatable governance.
📖 Quick Start Guide | 🚀 Accelerator Template | 📋 Full Workflow Docs | 🎯 Scenarios | 🧩 Azure MCP Server (GA) | 💰 Pricing MCP add-on
If you are implementing this workflow for a real environment, start with the accelerator template. It is pre-wired with the Agentic InfraOps workflow, automation, and repo structure so you can focus on requirements and governance, not setup.
- Template repository: azure-agentic-infraops-accelerator
- Use the accelerator when you want a clean workload repo that adopts this workflow.
- Use this repo when you want to review the agents, guardrails, and example outputs.
- The core enabler behind “agents with real Azure context”, not guesses
- RBAC-aware: tools operate within your existing Azure permissions model
- Accelerates day-0 to day-2 operations: discovery, validation, and troubleshooting workflows
- Reduces context switching: fewer portal loops, faster repeatable operations
- Broad coverage (40+ Azure service areas): platform, monitoring, governance, and more
- Works with Copilot Agent mode and multiple IDEs/clients
- This repo also includes a Pricing MCP add-on for cost-aware SKU decisions
Learn more:
- Azure MCP Server README: https://github.com/microsoft/mcp/blob/main/servers/Azure.Mcp.Server/README.md
- Microsoft Learn: https://learn.microsoft.com/azure/developer/azure-mcp-server/
For production adoption, start with the accelerator template above. For evaluation and learning, follow the steps below.
%%{init: {'theme':'base', 'themeVariables': {'primaryColor': '#0078D4', 'primaryTextColor': '#ffffff', 'primaryBorderColor': '#005A9E', 'lineColor': '#6B7280', 'secondaryColor': '#00B4AB', 'tertiaryColor': '#8957E5', 'background': '#ffffff', 'mainBkg': '#ffffff', 'nodeBorder': '#005A9E', 'clusterBkg': '#F3F4F6', 'titleColor': '#1F2937', 'edgeLabelBackground': '#ffffff'}}}%%
graph LR
P["🎯 Requirements<br/>Step 1"]:::plan --> A["🏛️ Architect<br/>Step 2"]:::architect
A --> D3["📊 Design Artifacts<br/>Step 3"]:::artifact
D3 --> B["📋 Bicep Plan<br/>Step 4"]:::bicep
B --> I["⚙️ Bicep Code<br/>Step 5"]:::bicep
I --> DEP["🚀 Deploy<br/>Step 6"]:::deploy
DEP --> D7["📄 Docs<br/>Step 7"]:::artifact
DEP -.->|validate| DX["🔍 Diagnose"]:::diagnose
AZMCP["🧩 Azure MCP Server (GA)"]:::mcp -.->|Azure context| A
AZMCP -.->|validation| B
AZMCP -.->|diagnostics| DX
MCP["💰 Pricing MCP (add-on)"]:::pricing -.->|costs| A
classDef plan fill:#8957E5,stroke:#6B46C1,color:#fff
classDef architect fill:#0078D4,stroke:#005A9E,color:#fff
classDef bicep fill:#00B4AB,stroke:#008F89,color:#fff
classDef pricing fill:#FF6B35,stroke:#E55A25,color:#fff
classDef mcp fill:#1D4ED8,stroke:#1E40AF,color:#fff
classDef artifact fill:#6B7280,stroke:#4B5563,color:#fff
classDef deploy fill:#10B981,stroke:#059669,color:#fff
classDef diagnose fill:#EF4444,stroke:#DC2626,color:#fff
Agent Legend
| Color | Phase | Description |
|---|---|---|
| 🟣 | Requirements | Gather and refine project requirements |
| 🔵 | Architecture | WAF assessment and design decisions |
| ⚫ | Design/Docs | Diagrams, ADRs, and documentation |
| 🟢 | Bicep | Implementation planning and code gen |
| 🔷 | MCP | Live Azure context and operational tools |
| 🟠 | Pricing | Real-time Azure cost estimation (MCP) |
| 🟩 | Deployment | Azure resource provisioning |
| 🔴 | Diagnose | Resource health validation & troubleshooting |
| Step | Phase | Agent(s) | Output |
|---|---|---|---|
| 1 | Requirements | @requirements |
01-* |
| 2 | Architecture | @architect 🧩 💰 |
02-* |
| 3 | Design | @diagram, @adr |
03-des-* |
| 4 | Planning | @bicep-plan 🧩 💰 |
04-* |
| 5 | Implementation | @bicep-code |
05-* |
| 6 | Deployment | @deploy |
06-* |
| 7 | Documentation | @docs |
07-* |
| — | Validation | @diagnose 🧩 |
08-* |
🧩 = Microsoft Azure MCP Server integration. 💰 = Pricing MCP add-on integration. Steps 3, 7 & Validation are optional. 9 agents total for the complete workflow.
Get up and running in 5 steps:
| Step | Action | Details |
|---|---|---|
| 1️⃣ | Install Prerequisites | Docker Desktop + VS Code + Copilot |
| 2️⃣ | Clone & Open | git clone https://github.com/jonathan-vella/azure-agentic-infraops.git then code azure-agentic-infraops |
| 3️⃣ | Open in Dev Container | Press F1 → "Dev Containers: Reopen in Container" (wait ~2 min) |
| 4️⃣ | Open Copilot Chat | Press Ctrl+Alt+I → Select Requirements from the agent picker dropdown |
| 5️⃣ | Try It | Type: Create a web app with Azure App Service and SQL Database |
Each agent asks for approval before proceeding. Say yes to continue, or provide feedback to refine.
📖 Full Quick Start Guide → (includes troubleshooting, demo scenarios, deployment instructions)
Agentic InfraOps is designed to be safe, repeatable, and governance-friendly.
- Custom Copilot agents codify the workflow:
see
.github/agents/ - Repository-wide instruction system enforces standards (Markdown, Bicep, agents):
see
.github/instructions/ - Drift-guard workflows protect templates and docs structure:
see
.github/workflows/ - AVM-first + CAF naming + required tags are built into the workflow defaults:
see
.github/agents/_shared/defaults.md
| Directory | Purpose |
|---|---|
.github/agents/ |
Agent definitions (9 agents for 7-step flow) |
.github/instructions/ |
Guardrails and standards (lint + consistency) |
.github/workflows/ |
Drift guard + automation workflows |
agent-output/ |
Generated artifacts per project |
.vscode/mcp.json |
Local MCP wiring for the Pricing MCP add-on |
mcp/azure-pricing-mcp/ |
💰 Pricing MCP add-on (real-time retail pricing) |
infra/bicep/ |
Generated Bicep templates |
docs/ |
Documentation, guides, diagrams |
scenarios/ |
8 hands-on learning scenarios |
Explore complete workflow outputs in agent-output/:
| Project | Description | Highlights |
|---|---|---|
| agent-testing | Agent validation framework | 16 resources, full 7-step + health validation |
| static-webapp | Static Web App with Functions | Production-ready SWA pattern |
| ecommerce | E-commerce platform | Multi-tier architecture |
8 hands-on scenarios from beginner to advanced (15-45 min each):
| Level | Topics |
|---|---|
| Beginner | Bicep baseline, diagrams as code |
| Intermediate | Documentation generation, service validation, troubleshooting, SBOM |
| Advanced | Full agentic workflow, async coding agent |
Efficiency multiplier: Reduce infrastructure development time by 60-90% while delivering Well-Architected, deploy-ready Azure infrastructure.
| Benefit | Details |
|---|---|
| AVM-First | Azure Verified Modules for policy-compliant deployments (ADR-003) |
| Time Savings | Quantified evidence: 45 min vs 18+ hours (time-savings-evidence) |
| Real Portfolios | See real projects built with agentic workflows (portfolio showcase) |
| Requirement | Details |
|---|---|
| VS Code | With GitHub Copilot extension |
| Dev Container | Docker Desktop or GitHub Codespaces |
| Azure subscription | For deployments (optional for learning) |
Included in Dev Container:
- ✅ Azure CLI with Bicep extension
- ✅ PowerShell 7+ and Python 3.10+
- ✅ All required VS Code extensions
- ✅ Pricing MCP add-on (auto-configured)
- ✅ Works best with the Azure MCP Server extension installed
Looking for a quick start? Check out the agentic InfraOps accelerator template: azure-agentic-infraops-accelerator.