sia.hackernoon.com

Hey HN! I built tf2cdk - a CLI that converts Terraform configurations to native AWS CDK code.

GitHub: https://github.com/jtaylortech/tf2cdk

Important: This is NOT cdktf

cdktf (CDK for Terraform):

Write CDK code that generates Terraform
Outputs HCL, uses Terraform state
For teams who want CDK's programming model with Terraform's engine

tf2cdk (this project):

Converts existing Terraform TO native CDK
Outputs CDK code, uses CloudFormation stacks
For teams migrating OFF Terraform entirely

Different problems. Not competitors.

The Problem

I've worked with multiple organizations migrating to AWS Control Tower. Every time, same issue:

Control Tower requires CloudFormation. Existing infrastructure is Terraform. Manual conversion takes 2-4 months for large codebases.

This also happens with:

Landing Zone Accelerator (LZA) - CloudFormation only
AWS Service Catalog - Native CloudFormation integration
Terraform licensing concerns - Teams want to eliminate Terraform entirely

How It Works

tf2cdk is a Python CLI that:

Parses HCL - Uses python-hcl2 to parse Terraform files
Maps resources - 30+ Terraform resources → CDK constructs
Generates code - Idiomatic TypeScript or Python
Validates - Type-safe with comprehensive error handling

tf2cdk convert ./terraform --output ./cdk-app --language typescript

All generated code lives in the output directory. You can inspect/modify it.

Example

Input (Terraform):

resource "aws_s3_bucket" "data" {
  bucket = "my-data-bucket"
  
  tags = {
    Environment = "production"
  }
}

resource "aws_s3_bucket_versioning" "data" {
  bucket = aws_s3_bucket.data.id
  versioning_configuration {
    status = "Enabled"
  }
}

Output (CDK TypeScript):

import * as cdk from 'aws-cdk-lib';
import * as s3 from 'aws-cdk-lib/aws-s3';

export class InfrastructureStack extends cdk.Stack {
  constructor(scope: cdk.App, id: string, props?: cdk.StackProps) {
    super(scope, id, props);

    const data = new s3.Bucket(this, 'Data', {
      bucketName: 'my-data-bucket',
      versioned: true,
      tags: { Environment: 'production' },
    });
  }
}

Notice how it:

Merges aws_s3_bucket_versioning into the bucket (CDK pattern)
Converts snake_case to camelCase
Generates clean, idiomatic code

Technical Details

Architecture:

Terraform HCL
    ↓ (python-hcl2)
Parsed Config
    ↓ (Resource Mapper)
CDK Model
    ↓ (Code Generator)
TypeScript/Python

Resource Mappings:

RESOURCE_MAPPINGS = {
    'aws_s3_bucket': {
        'cdk_class': 's3.Bucket',
        'module': 'aws-cdk-lib/aws-s3',
        'props': {'bucket': 'bucketName', 'tags': 'tags'}
    },
    # 30+ more...
}

Error Handling:

Validates source files exist
Catches HCL parse errors with helpful tips
Warns about unsupported resources
Shows conversion progress
Proper exit codes for automation

Testing:

28 tests (parser, converter, generator, CLI, edge cases)
Tests empty configs, unsupported resources, complex props
CLI integration tests for all commands
All tests passing

Supported Resources (30+)

Compute: EC2, Lambda, ECS, Auto Scaling Storage: S3, EBS, EFS Database: RDS, DynamoDB, ElastiCache Network: VPC, Subnet, Security Groups, Route53, ALB/NLB Security: IAM, KMS, Secrets Manager Monitoring: CloudWatch, SNS, SQS

Community contributions welcome for additional resources.

Special Patterns

# Control Tower account factory structure
tf2cdk convert ./terraform --pattern control-tower

# Landing Zone Accelerator format
tf2cdk convert ./terraform --pattern lza

# GovCloud compliance annotations
tf2cdk convert ./terraform --pattern govcloud

Why Open Source?

Terraform→CDK conversion is a one-time migration problem. You need it once, then you're done.

Making it free and open source:

Maximizes adoption
Gets community contributions (more resource mappings)
Helps the entire AWS community
Builds credibility

For large migrations needing implementation services, those are available separately. But the tool itself is free forever (MIT License).

Limitations

Complex Terraform expressions may need manual adjustment
Custom providers need manual mapping configuration
State migration is guided, not fully automated
Some Terraform-specific features (provisioners) have no CDK equivalent

Roadmap

v0.1.0 (current): 30+ resources, TypeScript/Python, basic patterns v0.2.0: 50+ resources, state import planning, more patterns v0.3.0: 100+ resources, module conversion, advanced expressions v1.0.0: Comprehensive resource coverage, GUI (optional)

Try It

# Homebrew
brew tap jtaylortech/tf2cdk
brew install tf2cdk

# pip
pip install tf2cdk

# Convert
tf2cdk convert ./terraform --output ./cdk-app --dry-run

Feedback Wanted

What Terraform resources are you using that aren't supported?
What patterns would help (multi-account, multi-region, etc.)?
What's your Terraform→CDK migration pain point?
Documentation gaps?

All code is on GitHub, MIT licensed. Issues and PRs welcome.

Related: If you're looking for CDK→Terraform (opposite direction), check out cdktf from HashiCorp. Different use case, both tools are useful.

Tf2cdk – Convert Terraform to AWS CDK (Not Cdktf)