Passkeys in Nuxt – The BEST Login UX
Five minutes of this removes hours of back-and-forth later.  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ 

Hi Adam,

Alex Garrett-Smith, Tech Education Lead at Unlearn.dev, has been building with AI long enough to know where the real time gets lost.

Not in the review. Not in the iteration. In the thirty seconds before the first prompt, most developers hand the agent a half-formed picture and expect a complete answer.

He had one of those moments on a rate-limiting feature. Simple ticket. He had built similar things before. He described what he wanted and generated.

The code came back clean. It compiled. It handled the happy path perfectly. It missed the distributed case entirely. Three instances running, clock skew between them, Redis going down under load. None of that was in the prompt. The agent had no way to know.

So now Alex does something before he writes any code.

He writes a spec for the agent. One paragraph. What the function accepts, what it must return, what happens on timeout, and what the caller assumes. Just a complete picture.

Then he asks the agent to break it before it builds anything.

"What have I missed in this spec? What assumptions am I making that could break in production?"


On that same rate-limiting feature, the agent came back with three production failure modes in under a minute. Things that would have surfaced in review or in an incident at 2 am. He fixed the spec. Added constraints: which packages, what to avoid, and how data should move. Then generated.

The output fit the codebase from the start. No back-and-forth on edge cases that were always going to be there.

This is the shift Unlearn is built around.

Alex recorded a short walkthrough of this exact workflow on a real feature that you can follow along step by step. All you have to do is request early access to Unlearn, and you'll get it in your inbox over the coming days.

Sign up in one click or learn more


Remember to secure your spot and get priority when limited founding seats open. Unlearn opens enrollment on March 31 for founding members for 48 hours only. Once that window closes, founding member access will be gone for good.

Best,
Maria

vueschool.io

Vue School is proudly part of BitterBrains – Making Developers’ Lives Easier

Become a BitterBrains Community Supporter here.

Courses  Vue.js Master Class  |  Workshops  |  Tutorials  |  Plans