GitHub: https://github.com/GBSOSS/skill-from-masters/

I've been building AI skills for a few months now. Recently I noticed something that changed how I approach skill creation entirely.

The Real Bottleneck

When I first started writing skills, I thought the hard part would be the technical stuff—getting the format right, making the instructions precise, handling edge cases.

I was wrong.

The actual bottleneck is knowing the best way to do the thing you're teaching the AI to do.

Think about it. If you're creating a skill for user interviews, the quality ceiling isn't determined by how well you write the SKILL.md file. It's determined by whether you know the right methodology for conducting user interviews in the first place.

Most people don't. I certainly didn't.

The Insight

Here's what I realized: almost every professional domain has masters who have already codified the best practices. They spent decades figuring out what works, then wrote books, gave talks, or shared frameworks.

For user interviews:

• Rob Fitzpatrick wrote "The Mom Test" — don't ask hypotheticals, ask about past behavior
• Steve Portigal wrote "Interviewing Users" — the definitive guide on rapport and questioning
• Clayton Christensen gave us Jobs to be Done — what job is the user hiring your product for?

For hiring:

• Laszlo Bock (Google's former HR head) systematized structured interviews in "Work Rules!"
• Geoff Smart's "Who" method has been used by thousands of companies
• Steve Jobs repeatedly emphasized "A players hire A players, B players hire C players"

For product decisions: