AI Pulse · Under the Radar

A GitHub skill for Claude called “I have ADHD” has quietly picked up 5,300 installs and 16,300 stars by doing something simple. It forces the model to answer first, use numbered lists, and skip the preamble. The creator built it for accessibility, but it spread fast because most people just prefer output that way.

The interesting part is not the tool itself. It’s that independent write-ups on Threads, Medium, and YouTube all converged on it within days of each other. No coordinated launch, no PR push. People found it, used it, and told others because it solved a friction point they did not know they could fix.

This is the second time in a week we have seen a formatting tweak go viral. Last week it was a prompt pattern that stripped out hedging language. This week it’s a skill that cuts straight to the answer. Both spread organically. Both address the same underlying complaint: default AI output is too verbose and too polite for real work.

If you run a team that uses AI tools daily, this pattern matters. Your people are already hunting for ways to make models more direct. They are copying prompts from Reddit, installing third-party skills, or just getting frustrated and moving slower than they should. You can wait for them to cobble together their own fixes, or you can bake better defaults into the tools you already use. The gap between stock AI and what your team actually needs is narrow, but it costs time every single day.

The original repo is public. If you want to see what people are responding to, the code is short and the comments are telling. This is not a feature request anymore. It’s table stakes.