Vibe-Coded-Disclosure

Philosophy

AI assisted code development is awesome (and sometimes infuriating).

It has made the present day arguably the greatest time since binary was invented to be a tech geek.

Equally, AI assisted code gen is very far from perfect. The task of getting AI to generate valid code is pertty Herculean: The process is challenged from many directions including the large context load posed by code-bases; the challenge of keeping LLMs with fixed training periods up to date with ever-evolving SDK and API syntax (etc).

But personally, I believe that AI code gen is a force for enormous good. I also believe that open source is a force for good. And that there is value in open-sourcing even buggy vibe-coded first iterations of an idea for other developers (robots and/or humans) to iterate upon, fork, and refine.

Vibe coders face some odd new moral quandaries: how do we not "plagiarise" non-human "collaborators" or authors like LLMs? What's the standard verbiage to say something like "FYI - this was written mostly by Claude and may be full of bugs. Use at your own risk!"

A Badge

Badges are good and much-loved ways to note things like these in tech docs.

One option is to add a simple "Vibe Coded" badge with Shields.io. Another is to note the model as a separate badge. Or do it together. I prefer to separate given that models are always evolving.