One of the downsides of using AI tools for development work or anything that requires predictable outcomes is that the same task isn't always performed the same way each time. 

Which is different than if I wrote a little bash or python script to handle a certain task. But isn't the promise and convenience of AI tooling is that it is easier and can be done with natural language?

If I want to spin up a new Craft project locally via Claude Code, I want to say: "Create a new Craft project called acme-corp" and Claude builds the new project exactly how I like.

Well, that's possible with Swamp and my new Swamp extension called Craft Starter. 

Swamp enables typed, composable models that allow you to define workflows that are invokable and repeatable by AI agents. You can use AI tools to build repeatable Swamp workflows and then invoke them from your favorite AI tool.

I also created another Swamp extension for provisioning a Hetzner server via a sentence in your AI agent harness.

These are just two experiences of what's possible and I created a lesson video showing how I use them.

➡️ Watch my new lesson to see both extensions in action.

CraftQuest is powered entirely by premium subscriptions. If you find the videos, articles, and livestreams helpful, please consider a premium membership to CraftQuest. Thank you!

More videos and tutorials soon,

-Ryan