I've avoided Terraform for years. Not because it's bad — it's excellent. But learning HCL, provider configs, state management, and module structures is a significant investment that I kept deprioritizing.
So I built nlops — describe your infrastructure in plain English, get production-ready Terraform back.
How a session looks
$ nlops init
Describe your infrastructure: A web app with a Postgres database, Redis cache,
and auto-scaling. Needs to stay under $150/month on AWS.
Generating Terraform...
Will create:
- 1 ECS Fargate service (web app, t3.small)
- 1 RDS Postgres instance (db.t3.micro, 20GB)
- 1 ElastiCache Redis node (cache.t3.micro)
- 1 Application Load Balancer
- Auto Scaling Group (min: 1, max: 3)
Estimated monthly cost: ~$127/month
Apply? [y/N]:
It generates real Terraform HCL. Not a wrapper. Not a platform. Standard .tf files you own forever.
Incremental updates
nlops update "add a CloudFront distribution in front of the load balancer"
nlops sends your existing Terraform + the change description to Claude, which returns a minimal diff. Only the new resources get added — your existing config stays untouched.
Drift monitoring
nlops watch
Runs terraform plan on a schedule. If actual infrastructure has drifted from your declared state, it alerts you and shows what changed.
The exit ramp
At any point:
nlops export
You get clean Terraform you can manage yourself, forever. No proprietary runtime, no lock-in. nlops is just a generator.
Why this matters
IaC is a solved problem technically. The barrier is accessibility. A developer who can describe what they need shouldn't also need to be an AWS specialist to deploy it safely.
nlops is for the developers who know what they want to build but don't want to spend a week learning provider-specific HCL before they can ship.
GitHub: https://github.com/LakshmiSravyaVedantham/nlops
Top comments (0)