Terraform Security Audit

Terraform Security Audit

Runs a 6-area security audit on Terraform codebases. Produces a pass/fail report per area.

When to Use

• Security review before production deployment
• Periodic infrastructure security audit
• New service setup validation
• Post-incident security hardening check

Process

1. IAM — Identity and Access Management

• OIDC used for CI/CD (no long-lived access keys)
• IRSA for EKS workloads (no node-level IAM)
• ECS task roles scoped per service (no shared roles)
• IAM policies follow least privilege
• No * actions on * resources
• No inline policies (use managed or customer policies)
• Assume role conditions include ExternalId or sts:SourceIdentity

# INSECURE — overly broad permissions
resource "aws_iam_policy" "bad" {
  policy = jsonencode({
    Statement = [{
      Effect   = "Allow"
      Action   = "*"
      Resource = "*"
    }]
  })
}

# SECURE — scoped to specific actions and resources
resource "aws_iam_policy" "good" {
  policy = jsonencode({
    Statement = [{
      Effect   = "Allow"
      Action   = ["s3:GetObject", "s3:PutObject"]
      Resource = "${aws_s3_bucket.assets.arn}/*"
    }]
  })
}

# SECURE — IRSA for EKS pods
module "irsa" {
  source = "git::https://github.com/{project}/terraform-modules.git//irsa?ref=v1.0.0"

  name              = "${local.name_prefix}-irsa"
  oidc_provider_arn = var.oidc_provider_arn
  namespace         = var.service
  service_account   = var.service
  policy_arns       = [aws_iam_policy.service.arn]
}

2. Network — VPC and Security Groups

• Databases in private subnets only
• No 0.0.0.0/0 ingress except ALB on 443
• Security groups use source_security_group_id, not CIDR
• Egress restricted where possible
• VPC flow logs enabled
• No public IPs on non-bastion instances

# INSECURE — database accessible from anywhere
resource "aws_security_group_rule" "rds_bad" {
  type        = "ingress"
  from_port   = 5432
  to_port     = 5432
  protocol    = "tcp"
  cidr_blocks = ["0.0.0.0/0"]
  security_group_id = aws_security_group.rds.id
}

# SECURE — database only from app security group
resource "aws_security_group_rule" "rds_good" {
  type                     = "ingress"
  from_port                = 5432
  to_port                  = 5432
  protocol                 = "tcp"
  source_security_group_id = aws_security_group.app.id
  security_group_id        = aws_security_group.rds.id
  description              = "PostgreSQL access from application"
}

3. Encryption — Data at Rest and in Transit

• S3: SSE enabled (KMS or AES-256)
• RDS: storage_encrypted = true
• RDS: ssl_enforcement via parameter group
• Redis: transit_encryption_enabled = true
• Redis: at_rest_encryption_enabled = true
• EBS volumes encrypted
• Terraform state bucket encrypted with SSE-KMS
• ALB uses TLS 1.2+ only

# INSECURE — no encryption
resource "aws_db_instance" "bad" {
  storage_encrypted = false  # default
}

# SECURE — encryption enabled
resource "aws_db_instance" "good" {
  storage_encrypted = true
  kms_key_id        = var.rds_kms_key_arn
}

# SECURE — Redis encryption
module "redis" {
  source = "git::https://github.com/{project}/terraform-modules.git//elasticache?ref=v1.0.0"

  transit_encryption = true
  at_rest_encryption = true
  auth_token         = var.redis_auth_token
}

4. Secrets — Secret Management

• No secrets in .tf files or committed .tfvars
• secrets.tfvars in .gitignore
• Sensitive variables marked sensitive = true
• Secrets injected via CI/CD environment variables
• No plaintext passwords in state (use sensitive output)
• git-secret-protector or pre-commit hooks block accidental commits

# INSECURE — password in code
resource "aws_db_instance" "bad" {
  password = "SuperSecret123!"
}

# SECURE — from variable, marked sensitive
variable "db_password" {
  description = "Database master password"
  type        = string
  sensitive   = true
}

resource "aws_db_instance" "good" {
  password = var.db_password
}

# SECURE — sensitive output
output "connection_string" {
  value     = "postgresql://${var.db_user}:${var.db_password}@${aws_db_instance.main.endpoint}/${var.db_name}"
  sensitive = true
}

5. Access — Cluster and Console Access

• EKS aws-auth ConfigMap restricts access to needed roles
• SSO used for console access (no IAM users with passwords)
• Bastion host in private subnet with Session Manager (no SSH keys)
• CloudTrail enabled for API audit logging
• MFA enforced on human accounts

# EKS access control
resource "kubernetes_config_map" "aws_auth" {
  metadata {
    name      = "aws-auth"
    namespace = "kube-system"
  }

  data = {
    mapRoles = yamlencode([
      {
        rolearn  = var.admin_role_arn
        username = "admin"
        groups   = ["system:masters"]
      },
      {
        rolearn  = var.node_role_arn
        username = "system:node:{{EC2PrivateDNSName}}"
        groups   = ["system:bootstrappers", "system:nodes"]
      }
    ])
  }
}

6. Compliance — Tags, Naming, and Backups

• All resources tagged: Project, Service, Environment, ManagedBy
• Naming follows {project}-{service}-{env} convention
• RDS automated backups enabled (retention >= 7 days, 30 for prod)
• S3 versioning enabled on data buckets
• DynamoDB point-in-time recovery enabled
• CloudWatch alarms on critical metrics
• Cost allocation tags configured

# CORRECT — default tags at provider level
provider "aws" {
  default_tags {
    tags = {
      Project     = var.project
      Service     = var.service
      Environment = var.env
      ManagedBy   = "terraform"
    }
  }
}

# CORRECT — backup retention
resource "aws_db_instance" "main" {
  backup_retention_period = var.env == "prod" ? 30 : 7
  backup_window           = "03:00-04:00"
}

Interaction Style

• Scans all .tf files in the codebase
• Checks every area — does not skip sections
• Highlights critical findings first (IAM wildcards, public access, missing encryption)
• Provides remediation code for each failing check

Rules

• Critical: IAM */*, public database access, unencrypted storage, secrets in code
• Warning: Missing tags, short backup retention, no flow logs
• Info: Missing descriptions, optional hardening not applied

Output

Produces a security audit report:

## Terraform Security Audit: {service}

### Overall: Pass | Fail

| Area       | Status   | Critical | Warnings | Info |
|------------|----------|----------|----------|------|
| IAM        | Pass     | 0        | 0        | 1    |
| Network    | Fail     | 1        | 0        | 0    |
| Encryption | Pass     | 0        | 1        | 0    |
| Secrets    | Pass     | 0        | 0        | 0    |
| Access     | Pass     | 0        | 0        | 1    |
| Compliance | Warning  | 0        | 2        | 0    |

### Critical Findings
- **[Network]** Security group `rds_main` allows ingress from 0.0.0.0/0 on port 5432
  - File: `modules/{service}/sg.tf:15`
  - Fix: Replace `cidr_blocks` with `source_security_group_id`

### Warnings
- **[Encryption]** Redis `at_rest_encryption` not enabled
  - File: `modules/{service}/redis.tf:8`
  - Fix: Add `at_rest_encryption = true`

### Remediation Priority
1. Fix critical findings before any deployment
2. Address warnings before production promotion
3. Info items for next sprint