Skip to main content
ATMOS.md3.0 KB
View on GitHub

Project Instructions

This file provides persistent context to the Atmos AI Assistant about this project.

Project Overview

This is a multi-region AWS infrastructure project demonstrating Atmos AI features. It contains:

  • VPC networking components across two AWS regions
  • Transit Gateway hub-spoke topology for cross-region connectivity
  • Network and production stacks per region

Architecture

  • Hub-spoke networking — us-east-1 hosts the Transit Gateway hub; us-west-2 connects via cross-region peering
  • Two regions — us-east-1 (primary) and us-west-2 (secondary)
  • Two stagesnetwork (shared networking) and prod (production workloads)

Stack Naming Convention

Stacks follow the pattern: {environment}-{stage}

  • ue1-network — Network stack in us-east-1
  • ue1-prod — Production stack in us-east-1
  • uw2-network — Network stack in us-west-2
  • uw2-prod — Production stack in us-west-2

Components

vpc

Virtual Private Cloud with CIDR allocation, availability zones, and NAT Gateways.

  • Network stacks use 10.1.0.0/16 (ue1) and 10.2.0.0/16 (uw2)
  • Production stacks use 10.10.0.0/16 (ue1) and 10.20.0.0/16 (uw2)

tgw/hub

Transit Gateway hub in us-east-1 network stack. Central routing for cross-region and cross-account connectivity.

tgw/attachment

Transit Gateway VPC attachment. Connects a VPC to the Transit Gateway. Present in all stacks.

tgw/cross-region-hub-connector

Cross-region Transit Gateway peering. Only in uw2-network, connects us-west-2 to the us-east-1 hub.

Component Dependencies

  • tgw/hub depends on vpc in the same stack
  • tgw/attachment depends on vpc (same stack) and tgw/hub (network stack)
  • tgw/cross-region-hub-connector depends on tgw/hub in ue1-network

Common Operations

Describe a component

atmos describe component vpc -s ue1-network

List all stacks

atmos list stacks

Validate configuration

atmos validate stacks

Check affected stacks

atmos describe affected

AI-Powered Analysis with --ai

The global --ai flag adds AI analysis to any command. The AI summarizes output on success and explains errors with fix instructions on failure.

Analyze command output

atmos terraform plan vpc -s ue1-network --ai
atmos describe component vpc -s ue1-prod --ai
atmos validate stacks --ai
atmos list stacks --ai

With domain-specific skills

# Single skill
atmos terraform plan vpc -s ue1-network --ai --skill atmos-terraform

# Multiple skills (comma-separated)
atmos terraform plan vpc -s ue1-prod --ai --skill atmos-terraform,atmos-stacks

# Multiple skills (repeated flag)
atmos terraform plan vpc -s ue1-prod --ai --skill atmos-terraform --skill atmos-stacks

Enable globally via environment variable

export ATMOS_AI=true
atmos terraform plan vpc -s ue1-network

# With skills via env var
ATMOS_AI=true ATMOS_SKILL=atmos-terraform,atmos-stacks atmos terraform plan vpc -s ue1-prod