Matt Pocock

I forget what it's called.

Matt Pocock, I think, made it.

I'm sure you know who that is.

So it became so popular, you do not need to install it.

with just like no skill installed.

But because it's so popular, the models know what that is.

So you don't actually have to install Matt Pocock's skill.

So it's like anything that becomes successful enough becomes like almost part of the model by definition.

No, it's a great, uh, thing to have.

I- it's, it actually reminds me of, uh, the Grill Me skill that was making some rounds, uh, a while back where, um, [lip smacks] uh, it was Matt Pocock I think that, that had it, or maybe it's, uh, one that he- similar to his, that was basically like, "Okay, if we're gonna start a new project, I want you to grill me on this thing.

I want you to, you know, keep asking questions, asking questions, ask everything until we're at a point where there are no more questions to be asked because we now have, you know, exhausted all the, you know, the, the answers that we need so that we can actually now execute on whatever the thing is that we're executing." And, you know, I've ha- I've, I've had several guests on the show that we talk about this idea of making a plan and how important the plan is, you know? And the plan isn't necessarily e- uh, precisely what's in your head.

There, uh, will be unknown unknowns, gaps that need to be, uh, identified and, and, uh, resolved, and those things, uh, really only come out with, uh, collaboration.

GitHub's top two trending repos are both agent skills libraries.

Matt Pocock slash skills, 2,207 stars in 24 hours, shell-based reusable skills for real engineers.

Dietrich Giebert slash ponytail, 1,539 stars, makes agents default to the laziest correct solution, both overtaking every coding agent in the trending list, right? On OpenRouter, number one by token volume is Hermes agent at 11.43 trillion tokens.

And Tencent open-sourced Team AI CLI, used internally since March.

A Git-based repo giving every agent on a team the same handbook.

Same problem Matt Pocock solves, now from a Tier 1 lab.

That's a tell.

The skills layer is where the votes are right now.

Um, her stuff has been, has been, you know, pushed around the internet lately about like how she's built these incredible verification loops and how they ship so much.

Uh, Ryan Carson has been talking a lot about this and recent stuff, Claire Vo, um, you know, Matt Pocock, uh, like there's a lot of stuff out there where you can go at, you can be technical and you can use your systems mindset to engineer these agentic loops.

And I, and I also think that, that there it's completely possible to build in, um, create like creative guidelines brand guidelines content guidelines look and it's not outsourcing completely there's there's still human direction like this is the goal this is the strategic goal or this is the strategic this is the training these are the examples um I

think there's a lot of that I can identify with you on that.

It's like, it's basically, it's just horrible.

And then literally Matt Pocock, who's like a really big person, like a really big skill writer, like stop hitting yourself.

And people like, we're increasing limits by 50%.

Be honest.

The pattern today? One word, skills.

Top two trending repos on GitHub are both agent skill collections, Matt Pocock's Skills for Real Engineers and Dietrich Gebert's Ponytail, both logging over a thousand new stars yesterday.

The community isn't just building agents, They're converging on reusable, composable skill files as the primary unit of agent infrastructure.

Honestly, that shift happened faster than I expected.

One of the things that helped me, and again, I think it ties into the pragmatic side of this.

I keep talking about Matt Pocock.

I think it's probably just like a meme at this point if you're going to do a drinking game.

It's a good one to, you know.