Infra As Code

Infra as Code — IaC Design Pipeline

An agent team collaborates to perform Terraform/Pulumi-based infrastructure design -> security -> cost optimization -> drift detection.

Execution Mode

Agent Team — 5 members communicate directly via SendMessage and cross-validate each other.

Agent Composition

Workflow

Phase 1: Preparation (Performed directly by Orchestrator)

1. Extract from user input:
   • Infrastructure Requirements: What service the infrastructure is for
   • Cloud Provider (optional): AWS / GCP / Azure
   • IaC Tool (optional): Terraform / Pulumi / OpenTofu
   • Constraints (optional): Budget, compliance, existing infrastructure
   • Existing Code (optional): Existing IaC code, architecture documents
2. Create _workspace/ directory at the project root
3. Organize input and save to _workspace/00_input.md
4. If existing files are available, copy them to _workspace/ and skip the corresponding Phase
5. Determine execution mode based on the scope of the request (see "Modes by Task Scale" below)

Phase 2: Team Assembly and Execution

Tasks 2a (security) and 2b (cost) can be executed in parallel.

Inter-team Communication Flow:

• architect completes -> delivers network, IAM, data stores to security; delivers resource specs and scaling to cost; delivers module structure and core resources to drift
• security completes -> delivers security policies and compliance checks to drift; delivers security cost items to cost
• cost completes -> delivers cost anomaly detection criteria to drift
• reviewer cross-validates all deliverables. Requests fixes for RED Must Fix items (up to 2 times)

Phase 3: Integration and Final Deliverables

1. Check all files in _workspace/
2. Verify all RED Must Fix items have been addressed
3. Report the final summary to the user

Modes by Task Scale

Leveraging Existing Files: If the user provides existing IaC code, architecture documents, etc., copy the files to the appropriate location in _workspace/ and skip the corresponding agent's step.

Data Transfer Protocol

Error Handling

Test Scenarios

Normal Flow

Prompt: "Design Terraform infrastructure on AWS for running a NestJS API server. Use ECS Fargate + RDS PostgreSQL + ElastiCache Redis, with dev/staging/prod environment separation." Expected Result:

• Design: VPC/subnet design, ECS/RDS/ElastiCache configuration, 3-environment module structure
• Security: Security group matrix, IAM roles, KMS encryption, Checkov policies
• Cost: Per-environment monthly cost estimates, Savings Plan suggestions, dev environment scheduling
• Drift: Security group/IAM immediate remediation, config drift alerts
• Review: Full consistency verification across all items

Existing Infrastructure Codification Flow

Prompt: "I want to convert infrastructure currently managed manually in the AWS console to Terraform" Expected Result:

• Import mode: Establish terraform import strategy
• Resource inventory, import command generation, state verification plan
• Include gradual migration roadmap

Error Flow

Prompt: "Create simple web server infrastructure" (no detailed requirements) Expected Result:

• Start design with basic configuration (VPC + EC2/ECS + ALB + RDS)
• Ask additional requirement questions (scale, DB, domain, etc.)
• Provide minimum configuration + expansion guide

Agent Extension Skills