Pangea

Scalable infrastructure management with Ruby DSL and template-level state isolation

Pangea is an automation-first infrastructure management tool that addresses key Terraform/OpenTofu scalability challenges through template-level state isolation, Ruby DSL compilation, and configuration-driven namespace management.

SaaS surface lives elsewhere. The hosted multi-tenant control plane is in pleme-io/pangea-api — OpenAPI-driven, with the typed Rust server, the saguão auth integration, multi-tenant Postgres isolation, and federation to per-region pangea-operator clusters. This repo remains the Ruby DSL authoring surface that customers (and the hosted control plane) consume.

Pangea compiles Ruby DSL templates to Terraform JSON and manages infrastructure with isolated workspaces per template, enabling infrastructure that scales with team size and complexity.

• Ruby 3.1+ (Ruby 3.3+ recommended)
• Terraform 1.5+ or OpenTofu 1.6+

gem install pangea

git clone https://github.com/drzln/pangea.git
cd pangea
bundle install
bundle exec rake install

Add to your Gemfile:

gem 'pangea'

Then run:

bundle install

Create Ruby files with template declarations:

# infrastructure.rb
template :web_server do
  provider :aws do
    region "us-east-1"
  end

  resource :aws_instance, :web do
    ami "ami-12345678"
    instance_type "t2.micro"

    tags do
      Name "WebServer"
    end
  end
end

template :database do
  provider :aws do
    region "us-east-1"
  end

  resource :aws_rds_instance, :main do
    engine "postgres"
    instance_class "db.t2.micro"
  end
end

Create pangea.yaml in your project root (see pangea.yaml.example for a complete example):

default_namespace: development

namespaces:
  development:
    state:
      type: local
      path: "terraform.tfstate"

  production:
    state:
      type: s3
      bucket: "terraform-state-prod"
      key: "pangea/terraform.tfstate"
      region: "us-east-1"
      dynamodb_table: "terraform-locks"

# Plan all templates in file (uses default namespace)
pangea plan infrastructure.rb

# Plan specific template
pangea plan infrastructure.rb --template web_server

# Plan with specific namespace
pangea plan infrastructure.rb --namespace production

# Show compiled Terraform JSON without running plan
pangea plan infrastructure.rb --show-compiled

# Show compiled JSON for specific template
pangea plan infrastructure.rb --template web_server --show-compiled

# Apply (auto-approves by default)
pangea apply infrastructure.rb --template web_server

# Apply with confirmation prompt (rare)
pangea apply infrastructure.rb --template web_server --no-auto-approve

# Destroy (auto-approves by default)
pangea destroy infrastructure.rb --template web_server

# Destroy with confirmation prompt (rare)
pangea destroy infrastructure.rb --template web_server --no-auto-approve

• Each template gets its own workspace and state file
• More granular than industry standard directory-based approaches
• Reduces blast radius and enables parallel development

• Auto-approval by default for streamlined CI/CD
• Non-interactive operation for automation workflows

• Type-safe resource functions with RBS definitions
• Compile-time validation using dry-struct
• Access to full Ruby ecosystem for complex logic

• Configuration-driven namespace management
• Support for local, S3, and custom backends
• Automatic state key generation prevents conflicts

• Templates are compiled to separate workspaces
• Each workspace manages its own Terraform state
• Namespaces configure shared backend settings
• No init command needed - initialization is automatic

• Reduced File Sprawl: Multiple templates in single files vs scattered directories
• Automatic Backend Management: No manual backend configuration per component
• Ruby DSL Power: Better abstraction than HCL for complex logic
• Code Reuse: Shared helper methods and logic within files

• Configuration Simplicity: Single YAML file vs multiple terragrunt.hcl files
• Template Isolation: Built-in template-level isolation vs manual workspace management
• Ruby Ecosystem: Access to full Ruby library ecosystem
• No DRY Complexity: Templates handle repetition naturally

• True State Isolation: Completely separate state files vs shared backend
• Security Boundaries: No cross-template state access
• Template Granularity: Template-specific configurations
• Operational Clarity: Template names match infrastructure components

See the examples/ directory for complete infrastructure templates:

• Simple Web Server - Basic single-template example
• Scalable Infrastructure - Multi-template web application
• Advanced ML Infrastructure - Complex healthcare ML platform

We welcome contributions! Please see CONTRIBUTING.md for details on:

• Development setup
• Running tests
• Submitting pull requests
• Code style guidelines

• GitHub Discussions - Questions and community
• Issues - Bug reports and feature requests
• Security Policy - Reporting security vulnerabilities

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

• Built on terraform-synthesizer for Ruby-to-Terraform compilation
• Inspired by the need for better infrastructure-as-code scalability
• Thanks to all contributors