GitHubHelp

devops4solutions/ai-assisted-devops-lab

★ 1Forks 0HCLGitHub ↗Compare

README

AI-Assisted DevOps Lab

This repository demonstrates the difference between:

  • 01-basic-prompt: plain Terraform implementation
  • 02-claude-md: Terraform implementation guided by a CLAUDE.md instruction file
  • 03-skils: AI-assisted workflow using CLAUDE.md + reusable Claude Skills

Quick Comparison

Area 01-basic-prompt 02-claude-md 03-skils
Main purpose Basic Terraform baseline Better Terraform structure through project instructions Scalable AI workflow with reusable skills
Project guidance Prompt-level only CLAUDE.md standards CLAUDE.md + .claude/skills/* skill packs
Terraform structure Simple flat setup Root split + module pattern Root split + module pattern + skill-driven consistency
Website source Not separated as app folder Infrastructure-focused Includes src/ static site files + terraform/ infra folder
Standardization Manual and ad-hoc Instruction-driven Instruction + reusable skill definitions
Reuse model Copy/edit files Reuse Terraform modules Reuse Terraform modules and Claude Skills
Team scale readiness Low Medium High

What Changed Across Stages

01 → 02

  1. Terraform becomes cleaner and more maintainable

    • Better separation of concerns (providers.tf, locals.tf, module usage).
    • Naming and tags become more standardized.

  2. Instruction-driven generation starts

    • CLAUDE.md sets expected project standards.

02 → 03

  1. From single instruction file to reusable skill system

    • Skills are added under .claude/skills/.
    • Example skill: .claude/skills/terraform-standards/SKILL.md.

  2. More generic CLAUDE.md

    • CLAUDE.md stays high-level.
    • Detailed behavior lives in skill files (better for token efficiency and reuse).

  3. Better workflow for larger teams/projects

    • Skills make standards repeatable across tasks and contributors.

Why CLAUDE.md Matters

CLAUDE.md acts like a project instruction contract for AI-assisted changes. It defines standards such as:

  • use reusable modules
  • avoid hardcoded environment/account values
  • use locals for derived naming
  • configure providers at root
  • enforce common tagging

Practical impact

When an AI assistant follows CLAUDE.md, generated code is typically:

  • more consistent across files
  • easier to maintain and review
  • closer to team standards
  • less likely to drift into one-off patterns

Without this instruction file, generated code may still work, but style and architecture can vary more between runs.

Why Skills Matter (in 03-skils)

Skills convert best practices into reusable, versioned instructions.

Practical benefits:

  • keep CLAUDE.md short and generic
  • reduce repeated long prompts
  • improve consistency across generated Terraform changes
  • make onboarding easier (new contributors follow the same standards)
  • scale AI-assisted delivery without losing governance

Note on .claude/settings.local.json

This file stores local permission decisions for Claude tooling (for example, allowing specific local commands). It is not your Terraform infrastructure code and does not affect AWS resources.

Suggested Learning Path

  1. Start with 01-basic-prompt to understand the minimal Terraform setup.
  2. Move to 02-claude-md to see how instruction-driven AI improves structure and standards.
  3. Move to 03-skils to learn how reusable skills make AI behavior more consistent and scalable.

Contributors

devops4solutions

Issues