AI Native Installer — AUTHOR

A friend asked if they could have a blog like mine. Self-hosted, cheap to run, no platform taking a cut.

Normally I’d give him a template repo, and a detailed README. Here’s how to set up AWS. Here’s how Pulumi works. Here’s the GitHub Actions OIDC configuration you need so you’re not storing AWS keys in CI. I’d have to sit with him one arvo and walk him through it.

Recently I was inspired by nanoclaw

git clone https://github.com/gavrielc/nanoclaw && cd nanoclaw && claude

Notice the && claude at the end. Instead of shipping with installer code, it comes with a CLAUDE.md that tells the AI how to step the user through configuration. And instead of having lots of plugins for different integrations, it has an AI skill for making custom integrations.

Not a complex README, an AI built in that walks them through every decision from first principles. So I took my blog stack (Astro, S3, CloudFront, Pulumi, GitHub Actions) and turned it into a template. The install is one line:

git clone https://github.com/lukerohde/aws-quill my-blog && cd my-blog && claude

Claude reads the CLAUDE.md, sees it’s a fresh clone, and offers to run /setup. From there it’s a conversation. It checks your dependencies, offers to install what’s missing, buys a domain through the AWS API if you need one, deploys two Pulumi stacks, creates the GitHub repo, sets the CI secret, and pushes the first commit. About $1/month to run on AWS.

The interesting part isn’t the infrastructure. That’s just plumbing I already had.

The interesting part is the CLAUDE.md and the skills — small markdown files that describe what to do and why. When something goes wrong, Claude finds a way around the problem. When the user needs to do something manually (like delegate nameservers at their registrar), Claude gives them the specific instructions for their specific registrar, then waits for confirmation.

It’s the difference between handing someone a recipe and cooking alongside them.

I keep thinking about what this means for platform teams.

Platform teams spend years building infrastructure patterns, starter kits, golden paths, internal developer platforms. The goal is always the same: let product teams move fast without reinventing infrastructure or making security mistakes. The reality is usually: months of work, for 10–1000 customers max, constant support tickets from people who got stuck.

The problem isn’t the infra patterns. It’s finding the right degree of modularity. Too prescriptive and a pattern doesn’t fit anyone. Too many modular building blocks, and there are too many dangerous loose wires.

An AI-guided template knows exactly what state you’re in. It can check whether your AWS credentials work before proceeding. It can read your existing config and fill in sensible defaults. It can refuse to do things that would break the security model — not because you programmed every failure mode, but because the AI understands the intent.

The guardrails are in the CLAUDE.md. The escape hatches are in the skills. You ship the platform as a template, with an AI guide, and the AI handles the installation and customisation. Far fewer support tickets, and standard patterns keep things maintainable.

Platform teams have always known that the hard part isn’t building the platform. It’s getting people to use it correctly. Now you can give them an AI that shows them how.

When testing aws-quill out, my jaw hit the floor when it brainstormed domain names with me, checked which were available, found the cost of each, and purchased the one I wanted for me. Wow!