AI on Kevin Quinn

My approach to Agent Skills: Occam's Razor

Tue, 21 Apr 2026 05:00:00 +0000

My current approach to Agent Skills is to not include them unless I feel the pain without them. The majority of the time I’m running 1-3 instances of vanilla Claude Code.

To be clear, I’m not against Skills usage. I pull in Skills that have clear value, like helping agents use specific tools more effectively (e.g. playwright-cli Skill). My only hold up with AI Augments¹ is that so few provide proof that they are better than using vanilla Claude Code.

Every Skill looks the same on paper

So far I’ve found many Skills to be a lot of hype, but not a lot to back up their claims. Without clear measures, how am I supposed to distinguish between your amazing Skill you put a ton of effort into refining, and one that ostensibly does the same thing, but was one-shot prompted by a 12-year-old trying to promote their YouTube channel?

To be fair, many Skills help with activities which are genuinely hard to measure. But it feels like if they had anything at all to show they’re better than other Skills (or vanilla Claude Code), they’d be screaming it from the rooftops and you wouldn’t be able to miss it. As it stands, even for a popular suite of skills like Superpowers: “An agentic skills framework & software development methodology that works” I struggle to find any concrete measures of WHY it’s better than plain vanilla tools. Ditto for basically every other Skill i’ve considered pulling in. If we can’t actually measure a Skill’s quality, are we just fooling ourselves by tweaking the tokens we use to arrive at the same place?

It feels a bit like we’re following the PKM trend, where people build up incredibly complex Obsidian/Notion workspaces to make them more productive, and at the end of it… they have a lot of artifacts of work, but not a lot of outcomes.

Long-term hypothesis: vanilla wins

I also subscribe to the hypothesis that the best Skills will eventually be eaten by the agent coding tools (like Claude Code), so I’ll end up getting the useful ones anyway. I could be wrong, but I believe we’ve already seen this happen with Claude Code with features like auto-memory, code execution, file tools like XLSX/PDF, scheduling routines, cloud agent runs, browser automation, as well as pre-built Skills which are bundled in without a user having to add them.

It’s such a great vetting ground for the teams at Anthropic, OpenAI, etc. to watch which ones are rising to the top in popularity and then figure out how to bake them into the core product. Easier to acquire someone else’s innovation than have all the ideas yourself.

Skills, commands, whatever the next hotness is. ↩︎

Is it really my project if AI wrote it?

Wed, 18 Mar 2026 05:00:00 +0000

I’m starting a new project, and I’m struggling with how much of it I should be writing by hand. I feel like a fraud if I use agentic engineering to develop it, but also that’s where the world is going. I think I’m afraid that people will view me as one of the people who vibe-codes a project in a couple prompts and then claims the work as their own. Part of this is deciding to push past the fear of other people’s opinions and just do it. I can choose to build this with care, while also learning the skills that seem to be the next abstraction layer for software engineers. I imagine it must have been pretty intimidating surrendering control to compilers and moving up the stack to higher level languages, so I can take some comfort that this dissonance is a feeling others have felt in the past & made it out the other side.

As a frustrating bonus, I don’t feel like I have time to allow myself to write a new project by hand. There is so much stuff being released, and the pace of AI is blistering, especially when I look at the prolific people who let down their internal guardrails and build like crazy. It feels like I HAVE to choose to develop this new project with AI so I can learn the skills. The more I see being released, I feel this impending rush that I’m not moving fast enough, that each new thing being created is something that I can’t be the one to make. I don’t feel smart enough to make really useful things, so if people nab all the low-hanging fruit, I’m donezo. That’s a zero-sum mentality on top of a scarcity mindset, neither of which feel like a good way to approach AI. We’re in a time of churn, there are no standardized best practices yet, which means there are no wrong ways to be learning right now. Everyone is still figuring things out, and I want that to fill me with hope. Whenever I feel this overwhelming dread, I want to try to toggle myself into curiosity mode instead.

I think I also need a mindset shift in what I consider effort. As it stands, if I didn’t write code by hand and have to struggle & flex my brain muscles, it doesn’t feel like I earned the right to say I built a project. Someone said it well when they compared AI to some sort of slot machine or TikTok, if I use them wrong I feel quite hollow. That’s on me though, because I don’t look at all the cool things Simon Willison is putting out and think “oh he didn’t really build them”. I look to him as one example of the prolific creators I want to emulate.

Update 3/26/2026: Excellent insights (from the creator of an agent framework, so it’s not just cope!) - Thoughts on slowing the fuck down | Mario Zechner . Big one: “And I would like to suggest that slowing the fuck down is the way to go. Give yourself time to think about what you’re actually building and why. Give yourself an opportunity to say, fuck no, we don’t need this. Set yourself limits on how much code you let the clanker generate per day, in line with your ability to actually review the code…..Learning to say no is a feature in itself.”

Being a docs nerd is a strength in the age of LLMs

Wed, 25 Feb 2026 06:00:00 +0000

As someone who struggles to remember anything, I don’t have the luxury of trusting my brain to keep track of it. I’ve been a “docs guy” on every team I’ve ever been on, mainly because I am a sucky colleague asking dumb questions 100 times if I don’t. In the age of LLMs, if you document your application and the history of decisions that have grown it well, you’ll get significantly more useful outputs when working with agents. Context engineering and all that jazz. For people who already have good habits around documentation, they get this bonus for free! Other people will need to learn & strengthen this muscle.

Applying to teams

In my experience, it’s always been a challenge to build a culture of documentation at work. People will agree that it sounds useful, but their lack of action speaks louder than their words (ironically the action would also be words, in this case). Even on other teams I worked with, I could see there was often just one hero doing the vast majority of writing up useful docs & keeping old ones up to date.

Fear not! For those docs nerds among us, we now have another arrow in the quiver of reasons to encourage people to contribute. Anecdotally & from skimming online, it seems people are more motivated to document if it helps a robot find answers, than if it would have helped your new-hire Craig. I guess f*ck Craig, but you’ll write up a million words to feed Claude? In this case, we gotta take what we can get.