I did it.

I figured out the perfect AI coding loop. It’s called the Finn Loop.

Since I started running it, my vibe coding output is up 100x and my own work is down by 95%.

And you can get the Finn Loop for free. But more on that at the end of this email.

Now, before I show you how it works, you need to understand the problem it solves:

Right now, most people vibe code in a single session.

They give the agent one idea, then babysit it through every step it takes.

But that workflow eats your entire day, and it only allows you to ship one feature at a time.

Whereas the Finn Loop distributes that same work across 3 sessions running 3 skills side by side each one with a single job:

1. /finn-spec turns your rough idea into a real spec
2. /finn-build picks that spec up and writes the code
3. /finn-review tests it and opens the PR

So let me walk you through all 3, one at a time.

1. Start every idea with /finn-spec

Type /finn-spec followed by your idea. But don’t overthink it, one rough sentence is plenty.

The skill will then ask you all the questions it needs to fully understand what you want, then file it as a real spec inside Linear.

In case you don’t know what Linear is, it’s a project management tool with a very generous free tier, and it’s where the spec sits so the other 2 skills can find it.

You need to get this part right, because nothing else in the Finn Loop will run until you do.

2. Run /finn-build in its own session

Open a second Claude Code or Codex session and leave a /finn-build loop running in it.

That way you never have to kick it off yourself. It’s already sitting there watching for work on its own.

The moment a spec lands in Linear, the /finn-build skill picks it up and, as the name says, builds it out for you.

When it’s finished, it updates the status of the issue in Linear on its own.

3. Run /finn-review in a third session

Now it’s time to run the /finn-review skill, which picks up every issue that’s been built out and reviews it.

It starts with security and optimization issues, then keeps going with:

• Testing the code in its own browser
• Laying out the test steps
• Taking screenshots
• And creating a PR

And finally, it puts the change in its own Vercel sandbox so you can click around a live version.

After it’s done, it pings you. I myself use Slack for this, but any messaging app you already use will work.

That one message carries the PR, the testing steps, an executive summary of the changes, and the link to the sandbox.

Which leaves you with just one thing to do:

Open it whenever you want, look at the change, and react with a rocket emoji. That signals to the review loop that the PR should be merged.

So the only manual steps in the Finn Loop are submitting the idea and verifying the final product at the end.

Everything in between is automated.

Sounds pretty good, right?

When you implement this Finn Loop, your day finally stops being this never-ending hand-holding process.

Instead, you wake up, make a cup of coffee, give your agent a list of ideas, and come back at night to review the changes.

You’re free to do whatever you want in between.

So where do you get the Finn Loop?

You can download it completely for free from this repo:

​Grab the full Finn Loop repo here.​

Block some time and try it today. I promise it will dramatically improve your workflow.

And if you’d rather not set it up alone, come do it with us.

I run a live session inside the Vibe Coding Academy every week, where 865+ members are shipping their own apps. Bring your first spec and we’ll build it together.

​Join the Vibe Coding Academy here.​

Keep shipping,

Alex