Terraform AVM Agent

An AI agent built with the Microsoft Agent Framework that generates Terraform code using Azure Verified Modules (AVM).

Features

• Service-Based Generation: Provide a list of Azure services and get a complete Terraform project
• Architecture Diagram Analysis: Upload an architecture diagram and the agent will identify services and generate code
• Interactive Chat Mode: Have a conversation with the agent to explore modules and generate code
• AVM Module Registry: Built-in knowledge of 40+ Azure Verified Modules
• Production-Ready Output: Generated code follows Terraform best practices

Installation

# Core CLI (list-modules, search, info, version, etc.)
pip install tf-avm-agent

# With AI agent support (chat, generate --interactive, diagram analysis)
pip install "tf-avm-agent[agent]"

# With API server
pip install "tf-avm-agent[api]"

# Everything
pip install "tf-avm-agent[all]"

# Development
pip install -e ".[dev]"

Extras

Prerequisites

• Python 3.10+
• An OpenAI API key or Azure OpenAI endpoint (for AI features)
• Terraform 1.9+ (for running generated code)

Set your API key:

export OPENAI_API_KEY="your-api-key"
# Or for Azure OpenAI
export AZURE_OPENAI_ENDPOINT="https://your-endpoint.openai.azure.com"
export AZURE_OPENAI_API_KEY="your-api-key"

Quick Start

Generate from Services List

# Generate Terraform for VM, storage, and key vault
tf-avm-agent generate -s "vm,storage,keyvault" -n "my-project" -o ./terraform

# Generate with specific location
tf-avm-agent generate -s "aks,acr,postgresql" -n "microservices" -l "westeurope" -o ./infra

Interactive Chat Mode

tf-avm-agent chat

Example conversation:

You: Generate Terraform for a web application with a database

Agent: I'll help you create a Terraform project for a web application with a database...
[Generates code using AVM modules for App Service, SQL Database, etc.]

List Available Modules

# List all modules
tf-avm-agent list-modules

# List by category
tf-avm-agent list-modules -c networking

# Get module info
tf-avm-agent info virtual_machine

Search for Modules

tf-avm-agent search "kubernetes"
tf-avm-agent search "database"

Python API

from tf_avm_agent import TerraformAVMAgent, generate_terraform

# Quick generation
result = generate_terraform(
    services=["virtual_machine", "storage_account", "key_vault"],
    project_name="my-app",
    output_dir="./terraform"
)

# Using the agent class
agent = TerraformAVMAgent()

# Generate from services
result = agent.generate_from_services(
    services=["aks", "acr", "postgresql"],
    project_name="microservices",
    location="westeurope"
)

# Interactive with AI
response = agent.run("Create Terraform for a secure web app with monitoring")

# Analyze architecture diagram
response = agent.analyze_diagram(
    image_path="./architecture.png",
    project_name="cloud-app"
)

Available Modules

The agent includes knowledge of 40+ Azure Verified Modules across categories:

Generated Project Structure

my-project/
├── providers.tf      # Provider configuration
├── variables.tf      # Input variables
├── main.tf           # Resource definitions using AVM modules
├── outputs.tf        # Output values
├── terraform.tfvars.example  # Example variable values
├── .gitignore        # Git ignore file
└── README.md         # Project documentation

Example Generated Code

# main.tf
module "virtual-machine" {
  source  = "Azure/avm-res-compute-virtualmachine/azurerm"
  version = "0.18.0"

  name                       = "virtual-machine"
  resource_group_name        = azurerm_resource_group.main.name
  location                   = azurerm_resource_group.main.location
  virtualmachine_sku_size    = "Standard_D2s_v3"
  virtualmachine_os_type     = "Linux"
}

module "storage-account" {
  source  = "Azure/avm-res-storage-storageaccount/azurerm"
  version = "0.5.0"

  name                     = "storage-account"
  resource_group_name      = azurerm_resource_group.main.name
  location                 = azurerm_resource_group.main.location
  account_tier             = "Standard"
  account_replication_type = "LRS"
}

Configuration

Environment Variables

CLI Options

tf-avm-agent --help
tf-avm-agent generate --help
tf-avm-agent chat --help

Architecture

flowchart TB
    subgraph Agent["Terraform AVM Agent"]
        direction TB
        CLI["CLI Interface<br/>(Typer + Rich)"]

        subgraph Core["Agent Core"]
            direction TB
            MAF["Microsoft Agent Framework"]
            DA["Diagram Analyzer"]
            AVM["AVM Module Lookup"]
            TG["Terraform Generator"]
            MAF --> DA
            MAF --> AVM
            MAF --> TG
        end

        Registry["AVM Module Registry<br/>(105 modules)"]

        CLI --> Core
        Core --> Registry
    end

Development

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run linting
ruff check src/

# Type checking
mypy src/

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

🤖 Fun fact: This project was built with significant contributions from GitHub Copilot and Claude. They wrote most of the code while the human mostly typed "fix this" and "that doesn't work". The AI agents are still waiting for their contributor badges.

License

MIT License - see LICENSE file for details.

Resources

• Azure Verified Modules
• Microsoft Agent Framework
• Terraform Azure Provider