Infrastructure As Code Patterns

Infrastructure-as-Code Patterns

Define and manage infrastructure through code for reproducibility, version control, and automation.

Context

You are building infrastructure-as-code practices. Choose tools, design module structure, plan testing and validation, implement drift detection. Read current infrastructure, team programming skills, and cloud platform.

Domain Context

Based on IaC best practices (Terraform, CloudFormation, Pulumi):

• Declarative vs Imperative: Declare desired state (Terraform) vs describe steps (CloudFormation). Declarative is more maintainable.
• Modularity: Reusable modules (VPC, database, load balancer); parameterized for different environments
• State Management: Terraform state tracks infrastructure; must be versioned and backed up
• Testing: Unit tests (validate syntax), integration tests (deploy to staging), policy tests (compliance)
• Drift Detection: Find manual changes not in code; reconcile or remediate

Instructions

1. Choose Tool: Terraform (multi-cloud, popular, large ecosystem). CloudFormation (AWS-native, less learning). Pulumi (programming language, more flexible).

1. Design Module Structure: Root modules per environment (dev, staging, prod). Shared modules for common patterns (VPC, RDS cluster, service). Versioned modules in registry.

1. Plan State Management: Store state in remote backend (S3 + DynamoDB for Terraform). Enable versioning and locking. Restrict access (IAM for AWS state). Never commit state to Git.

1. Implement Testing: Validate syntax (terraform fmt). Unit tests (mock resources). Integration tests (deploy to ephemeral environment, test behavior). Destroy after test.

1. Detect and Handle Drift: Use terraform plan to detect changes not in code. Use policy as code (Sentinel, OPA) to enforce compliance. Periodically reconcile drift.

Anti-Patterns

• One Monolithic Module: All infrastructure in single file. Result: hard to understand, reuse, test. Guard: Break into modules; each module responsible for logical unit.
• State Committed to Git: Store state in version control. Result: leaks credentials, creates merge conflicts. Guard: Remote state backend; ignore .tfstate in Git.
• No Testing: Deploy to production first time. Result: broken infrastructure. Guard: Test in ephemeral environment; validate before production.
• Ignoring State Locking: Multiple people apply changes simultaneously. Result: state corruption. Guard: Use remote backend with locking; enforce plan review before apply.

Further Reading

• Terraform: Up and Running by Yevgeniy Brikman — foundational IaC patterns
• Infrastructure as Code by Kief Morris — IaC principles and practices
• Pulumi Documentation — multi-language IaC approach

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: sethdford
• Source: sethdford/claude-skills
• License: MIT

Install and usage instructions live in the source repository linked above.