Big Idea

The argument in one line.

Skills v1.2.0 shows a consistent design principle: give the agent a controlled vocabulary and structure for judgment calls, but hand irreversible or credential-touching steps to a deterministic script a human runs by hand.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You already use Claude Code, Codex, or a similar agent harness with a custom skills or AGENTS.md setup and want current patterns for keeping agent output disciplined.
You maintain your own reusable skill or prompt library and want ideas for versioning, packaging, and documenting it for other people to install.
You've been frustrated by an AI model producing verbose or confusing output and want a concrete fix pattern, not just 'try a different prompt.'
You're building a setup or provisioning flow and are unsure whether to hand it to an agent or keep a human driving it manually.

SKIP IF…

You've never used an AI coding agent or a skills/plugin system — terms like SKILL.md, AGENTS.md, and openai.yaml won't mean anything yet.
You're looking for a general AI-coding tutorial — this is a changelog and demo video for one person's specific tool, not a from-scratch course.

TL;DR

The full version, fast.

Skills v1.2.0 adds a real documentation site, official Claude Code marketplace packaging, and Codex compatibility via per-skill config files, then ships five skills: Wait What forces a model into plain, project-specific language when its output turns to jargon; Grill Me now asks dependency-ordered rounds of questions instead of one at a time; Writing for Agents makes skill files and AGENTS.md/CLAUDE.md more concise; Wizard turns tedious infrastructure provisioning into a deterministic, human-driven bash walkthrough that never touches an AI; and To Questionnaire exports unresolved decisions into a document for people outside the AI conversation. The throughline: automate repetition, keep humans or scripts in charge of anything irreversible.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:18

01 · Cold open

Announces v1.2.0 and notes the skills repo has passed 204,000 GitHub stars, framing the release as a mature project needing real docs.

00:18 – 01:18

02 · New documentation site

Tours the new aihero.dev/skills site: skill groupings, a full reference panel, an FAQ sourced from a personal wiki, and a linked AI-coding dictionary.

01:18 – 01:43

03 · Claude Code marketplace integration

Skills are now installable as a read-only bundle from Claude Code's official plugin marketplace, with automatic updates.

01:43 – 02:32

04 · Codex compatibility improvements

User-invoked (hidden-until-called) skills didn't work in Codex; every skill now ships a sidecar openai.yaml with allow_implicit_invocation: false to fix that.

02:32 – 04:01

05 · Wait What skill for clarity

A new skill for when Opus-family output turns verbose and jargon-heavy: it enforces Simplified Technical English and grounds the model in the project's own vocabulary.

04:01 – 06:27

06 · Grill Me multi-question rounds

Grill Me no longer asks one question per turn; it models questions as a dependency graph and asks only currently-answerable ones, in rounds.

06:27 – 07:27

07 · Writing for agents skill

The old 'writing great skills' skill is generalized into Writing for Agents, for making any agent-facing document (skills, AGENTS.md, CLAUDE.md) concise and predictable.

07:27 – 08:58

08 · Wizard skill for provisioning

A new skill that generates a deterministic, interactive bash wizard for infrastructure steps only a human can do — logins, API keys, cloud consoles — never handed to an agent.

08:58 – 10:28

09 · To Questionnaire for collaboration

Exports the unresolved decisions from a grilling session into a document someone outside the conversation (e.g. a spouse or stakeholder) can fill in and hand back.

10:28 – 11:37

10 · Wrap up and course announcement

Points to the full changelog on GitHub and announces a self-paced AI Coding Crash Course waitlist.

Atomic Insights

Lines worth screenshotting.

A skills repo with 204,000+ GitHub stars gets a dedicated documentation site instead of just a bigger README, including a personal glossary of AI-coding terms.
Claude Code's official plugin marketplace installs a skill bundle read-only, and every future update is pulled down automatically with zero manual re-install.
Codex doesn't support hiding a skill from the agent's context until explicitly invoked, so v1.2.0 ships a sidecar openai.yaml per skill to fake that behavior.
The real fix for a verbose, jargon-heavy model isn't telling it to 'use simple language' — it's grounding it in the specific vocabulary your own project already defines.
A skill called Wait What exists purely to catch unreadable model output and force an on-demand plain-language rewrite.
One-question-at-a-time interviews get tedious once only trivial yes/no confirmations are left; batching questions into rounds keeps the pace up.
Modeling interview questions as a dependency graph lets an agent ask only what's currently answerable, in waves, without breaking logical order.
A skill for improving OTHER agent-facing documents (skill files, AGENTS.md, CLAUDE.md) is worth maintaining separately from skills that perform tasks.
Pulling sprawling instructions out of a single AGENTS.md and into individual skills avoids front-loading the agent's context with everything at once.
For steps that touch real credentials or cloud consoles, a deterministic bash script that walks a human through each click beats letting an agent attempt it via computer-use.
The author deliberately avoided giving an agent computer-use access to provision infrastructure, even though it was technically possible, to keep full control over irreversible steps.
When a decision depends on someone outside the current AI conversation, exporting the open questions into a plain shareable document is faster than looping them into the session.
A workaround skill built to patch a collaboration gap (To Questionnaire) is explicitly designed to become unnecessary once agents get native multiplayer support.
The author's read on AI coding practice: the fundamentals have barely shifted since around December of the previous year, which is why a self-paced course now makes sense over a cohort.

Takeaway

Keep humans in control of the risky steps

SKILL DESIGN LESSONS

Across five new or updated skills, the pattern repeats: automate the tedious parts, but keep a human or a deterministic script in charge of anything irreversible or judgment-based.

02New documentation site

Treating a widely-used open-source repo's docs as a real product — with a glossary of terms and a sourced FAQ — turns a README into a teaching tool, not just a reference.
Linking every technical term in your docs to a plain definition helps readers use your tools correctly instead of guessing what jargon means.

03Claude Code marketplace integration

Shipping through an official plugin marketplace as a read-only bundle means users get every update automatically, with zero manual re-install steps.

04Codex compatibility improvements

A feature that works in one agent harness doesn't automatically work in another — always check per-harness support before relying on it.
Shipping a small sidecar config file per unit is a workable way to port a harness-specific behavior elsewhere without rewriting the core logic.

05Wait What skill for clarity

Forcing a model to use a controlled-language standard is a blunt but effective lever against verbose, jargon-heavy output.
Telling a model to 'use simple language' is a weaker fix than grounding it in the specific vocabulary your own project already defines.

06Grill Me multi-question rounds

A rigid one-question-per-turn interview format gets slow once only trivial confirmations are left — batch the easy ones together.
Modeling interview questions as a dependency graph lets you answer only what's currently unblocked, in waves, without breaking logical order.

07Writing for agents skill

A tool that improves OTHER documents is worth maintaining separately from tools that perform tasks — its only job is making agent-facing text concise and predictable.
Pulling sprawling instructions out of one giant config file and into smaller, targeted pieces avoids overloading a system's context with everything at once.

08Wizard skill for provisioning

For steps that touch real credentials or cloud consoles, a deterministic script that walks a human through each click beats letting an agent attempt it autonomously.
A provisioning flow that shows progress, opens the right URLs automatically, and saves secrets where they belong turns a dreaded manual process into something closer to painless.

09To Questionnaire for collaboration

When a decision depends on someone outside the current conversation, exporting the open questions into a plain shareable document is faster than trying to loop them into the tool.
A workaround built to patch a collaboration gap should be treated as temporary — the real fix is the underlying platform gaining native support for that use case.

Glossary

Terms worth knowing.

Grill Me: A skill that interviews the user with questions to sharpen a plan before an agent acts on it, now redesigned to ask dependency-ordered rounds instead of one question per turn.
Wait What: A skill invoked when a model's output turns verbose or confusing; it forces a rewrite in plain, declarative language grounded in the project's own vocabulary.
Writing for Agents: A skill and reference file for making any document an agent reads — skill files, AGENTS.md, CLAUDE.md — more concise and predictable.
Wizard skill: A skill that generates an interactive bash script walking a human, step by step, through infrastructure or credential steps only they can perform — never handed to an AI agent.
To Questionnaire: A skill that exports the unresolved decisions from an interview-style session into a markdown document someone outside the conversation can fill in.
Simplified Technical English (STE-100): A controlled-language standard that requires short, plain, declarative sentences; used here as an instruction to stop a model from writing dense jargon.
allow_implicit_invocation: false: A configuration flag that hides a skill from the agent's context window until the user explicitly invokes it, rather than letting the model decide to use it on its own.
ubiquitous language: The specific vocabulary a project has already defined for itself (e.g. in a context.md file); grounding a model in this language is presented as the real cure for generic-sounding output.

Resources

Things they pointed at.

00:21linkaihero.dev/skills documentation site ↗

10:57productAI Coding Crash Course (waitlist) ↗

00:00linkAI Hero Discord ↗

00:00channelMatt Pocock on Twitter/X ↗

Quotables

Lines you could clip.

02:38

“Every time I interact with Opus, it just goes right over my head. It's incredibly verbose.”

names a specific, widely-relatable model-behavior complaint→ TikTok hook↗ Tweet quote

04:38

“If you think about it, these questions are really like a graph.”

a clean mental model for a UX problem, standalone→ IG reel cold open↗ Tweet quote

08:50

“This takes provisioning services from incredibly painful to weirdly joyful.”

punchy before/after claim about a tedious task→ newsletter pull-quote↗ Tweet quote

10:57

“There's not actually that much has changed since about December last year.”

contrarian claim about the pace of AI-coding change→ TikTok hook↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

metaphoranalogy

Skills. Skills. Skills, folks.

We have yet more skills for you. This is version 1.2 o which adds a bunch of improvements to my already extremely popular engineering skills.

The skills repo is now, I think, the the twenty fourth most starred repo of all time on GitHub up to 204 k stars. And as a mature project, we need some proper documentation. And so we have it.

This is at a ihero.dev/skills. And I've been working with my team to produce this absolutely wonderful site.

The best way to navigate this is by going to the main page forward slash skills and then looking at the kind of groupings that we've got here. For instance, we can see here the main flow where you start with grill with docs, then go to to spec, then to tickets, implement, and then code review. In the left hand panel here, you've got a full reference of every single skill and what they do.

It used to be my actual, uh, full time job to write documentation, so I was pretty happy to go back to it. There's even a list of common questions which are actually sourced from a personal wiki that I maintain of all the questions people ask me.

These skills are also linked to an AI coding dictionary that I've been putting together. For instance, this little ticket link here is the dictionary definition for what I think of as a ticket. So these docs not only can you use them to learn my skills better, you can also use them just to learn about AI coding and understand how things work at a deep level.

The second big change is we are now an official part of the Claude code official plug in marketplace. What that means is you can open Claude Code. You can say plug in like this.

You can search for Matt Pocock here or Matt Pocock skills and then you can install it right from here. This means you end up with a read only bundle. There's no extra steps here and any updates I make to the skills will be automatically pulled down to you.

That sounds pretty good if you're a Claude code user, but what if you're a codex user? What's in this release for you? Well, I now include open a I dot yaml files for every single skill, which means that it should properly work out of the box with UIs like Codecs UI.

And also, it means that this allow implicit invocation false travels over to Codecs. Before, we had model invoked skills and user invoked skills.

And the benefit of user invoked skills, which we can see inside the main skilled r m d file, is these are hidden from the agent's context window until you invoke them. That works for Claude code. It works for py.

It works for a couple of other harnesses. But it doesn't work for codecs. I didn't quite realize that.

And so now we ship these sidecar files with every single skill to mean that we have this allow implicit invocation false, so that works out of the box with codecs too. The next thing I wanted to ship a skill for is the Opus, especially Opus five, is talking garbage at the moment.

I don't understand what it is. I think it might just be a new model thing. But for some reason, every time I interact with Opus, I it just goes right over my head.

It's incredibly verbose. It uses really weird LLM phrases and I just find it really really hard to read and a lot of people feel the same.

The honest options are to accept it and reword the AC or to persist the digest along the export so a resumed unchanged publish can skip the read to flagging rather than picking. What does that mean? It is genuinely load bearing though, so that's how you know it matters.

I've tried all sorts of ways to fix it such as output styles or adding things to my agents.md. But what I realized I needed was just a skill to say, wait, what did you just say? And so the skill is called wait, What?

It's a very, very simple skill that just does a couple of things. First, it says use this specific standard ASD STE 100 simplified technical English, which is essentially just a leading word to the agent to say use very simple language.

Speak in clear declarative sentences. And secondly, it also tells it to ground itself in the ubiquitous language in context dot m d.

The real cure for verbosity is not to tell it to use simple language, although we are doing a bit of that. It's to tell it to use your language, the stuff that you have come up with in Grill with Docs. You use this skill whenever the agent just creates some random garbage and you've no idea what they just said.

You just say, wait, what? And it should reply with a much better alternative. The next thing that's changed is a pretty big one.

It's an update to my most popular skill, which is grill me. Previously, grill me used to grill you one question at a time. In other words, you get a question and then you would respond, and then another question then you would respond.

So one question per turn. However, this has a horrible failure mode which is when you get to the end of a grilling session and most of the hard stuff is already done, you just have a bunch of easy questions one per turn that you're just basically saying, yeah. That sounds good.

Yeah. That sounds good too. This is incredibly frustrating and it feels dead slow.

So for a while, I've wanted to find a way to speed this up. The thing that makes most sense to me is to have multiple questions per turn. So if you have lots of easy ones, you can just say yes.

Yes. Yes. Yes.

Yes. But what happens when the answer to question three depends on the answer in question one? Before, you never had that problem because you would just answer them one at a time.

So if a question depended on an earlier question, then that one had already been answered. But in this new setup, surely that's more confusing because you're gonna be answering erroneous questions or questions at the wrong time or questions need that to be answered before other questions. How does that gonna work?

Well, if you think about it, these questions are really like a graph. There might be one critical question at the very start that needs answering first that opens up a whole raft of other questions, and maybe they lead on to other questions themselves. And so now, grill.me is designed to follow this graph to answer only the questions that are available now, but to ask them in rounds.

And so you get the first round of questions, which might be as little as one critical question, Then you go to the next round with the questions that opens up, and it's always pushing you as fast as possible down the frontier of questions. And this is what it looks like.

You can see I'm at round one here, and we have q one, and then we get this little recommendation here. I've decided to use I know I know you probably don't like it but I love it.

Emojis here for a little pop of color. This makes it super easy to navigate with my eyes. I can see, okay, this is q one, and then I can see the recommended answer.

Q two, see the recommended answer. And because I'm using dictation, I just get to blast out. Okay.

Q one, I agree. Q two, I agree. Q three, we need something to change there.

Q four, we need something to change. And then further down, we can see round two came in and the same setup. So we answered the first round of questions, and then we got another round of questions.

And we kept pushing the frontier back. The next big change is that I've modified my writing great skills skill because I noticed that I was using it for stuff other than skills. I was using it for anything that an agent read to make the agent better, to make the doc more concise, and to make the output a bit more predictable.

And so now we have writing for agents where you can modify it for agents.md or claw.md or when you're creating editing skills and it has some extra skill mechanics stapled on inside this reference file.

So you can use it for writing skills. You can use it crucially for taking your agents.md, pulling it out into skills to stop that horrible front loading.

So overall, I find myself using this skill so so often whenever I need to write any agent configuration and I want it to read well, I want the agent to perform well with it, writing for agents is the right call. And it's also model invocable as well and it's a really clean description here.

So it should pull it in whenever it's modifying agents.md which is just great. The next one is a new skill, one that I've been keeping under wraps for a little while.

This is the wizard skill. I found myself recently having to provision a bunch of infrastructure and walk through a bunch of AWS stuff and I just hated it.

And I thought surely there's a way that the agent can help me here. And so I created this wizard skill that generates an interactive bash wizard that walks the human through steps only they can perform. In other words, I probably could have got the agent to, okay, go into AWS and provision the stuff with computer use or something, but it just felt pretty icky.

I wanted to have control over all of it, but I wanted it to be as easy as possible. And here is what one looks like. This was for I think migrating to a remote box setup.

So you just kick it off and then you say ready to start? Yes, please. And then it sends me off to exactly the page I need to go.

That opened the script by the way. I didn't need to do anything there. Sends me off, guts me to log in, gets me to change the exact thing we need to.

I can then paste in my API keys and remember this is a deterministic script so nothing's touching an agent here. It's not sending it off to Anthropic or anything.

So I paste it in here. It saves it into the files it needs to. It's even saving them into GitHub secrets if it needs to.

And it walks through these separate stages with me, keeping going with this nice little UI until we end up with what we need. I found that this takes provisioning services from incredibly painful to weirdly joyful because I just know how long it took me before.

So wizard has become an essential part of my skill set. I had it in in progress for a couple of months and I just think, okay, it's time for you guys to see it. The next one is certainly the most boring skill name that I've ever produced, but I found myself really, really needing it and I realized it would be useful for more than what I had just used it for.

It is called two questionnaire and its purpose is that you take the decisions inside a grilling session and you pull them out into a document that you can walk through with someone else. I was doing a wayfinder session to try to build a garden office in my back garden, And I realized that the agent was grilling me, but the person I really needed to speak to about it was my wife because my wife, obviously, was gonna use this space too, and so we're going to design it together.

And so I got the agent to take the stuff it was gonna ask in the grilling session and turn it into a markdown document. I then put that into a Google doc. I sent that over to my wife and we walked through it together, and then we pulled the answers back into the agent.

Now this is a skill that I hope someday to delete because it's sort of like a patch for the fact that agents are kind of hard to collaborate with at the moment. Some teams who have like, uh, what I consider a really good setup is they have the agent in Slack and they have these conversations they can then tag the agent in.

So they can collaborate together, answer questions together, especially with stakeholders, And then the agent can actually pile in and go, okay. I'll implement it based on that.

But a lot of people are not in that world right now. A lot of people are not AI native. Maybe they don't even have Slack or Microsoft Teams or something.

And so actually having a document that you can pull out, give to someone, they can comment on it on Google Docs or whatever is actually really useful. And so that folks is basically it. We've got a tranche of new skills.

I'm especially looking forward to your reactions to wait what because I think it will actually save your ass a couple of times when Opus is doing strange things. There were a few things that were just a bit too small to mention here and so you can go to v 1.2 o in the GitHub releases to take a look at the full change log.

And if you dig in this stuff, then I'm putting out an AI coding crash course, which is pretty different to the paid courses that I've put out before. Instead of being a cohort where you do it over a couple of weeks and everyone's doing it together, this is gonna be a cheaper course that's available throughout the year. My personal opinion about AI coding is that it's kind of settling down a little bit.

There's not actually that much has changed since about December last year. So And I now feel confident enough that I can put out a self paced course that hopefully won't change too much even as things develop. I'm hoping to give you a stable foundation from which you can work and actually ship amazing stuff.

And the link to this is down below and you can join the wait list. It should be coming out only in a couple of weeks, I think. So thank you so much for watching.

Enjoy the new skills and I will see you very soon.

The Hook

The bait, then the rug-pull.

Matt Pocock walks through Skills v1.2.0 — a proper documentation site, official Claude Code marketplace packaging, and five new or reworked skills, from a fix for garbled model output to a bash wizard that keeps AI agents away from your cloud credentials.

Frameworks

Named ideas worth stealing.

00:00list

Skills v1.2.0 — five new/updated skills

Wait What — fixes verbose/jargon-heavy model output
Grill Me — now asks dependency-ordered rounds of questions
Writing for Agents — makes skill files, AGENTS.md, CLAUDE.md concise
Wizard — deterministic bash walkthrough for infra provisioning
To Questionnaire — exports open decisions into a shareable doc

The full skill line-up shipped in this release, alongside documentation-site and marketplace/Codex packaging changes.

Steal forany personal skill or prompt library that needs versioning and distribution conventions

CTA Breakdown

How they asked for the click.

VERBAL ASK

10:57product

“The link to this is down below and you can join the wait list.”

Soft CTA placed only in the final ~40 seconds, after all the value is delivered; frames a self-paced course as a calmer, cheaper alternative to prior cohort-based paid courses.

MENTIONED ON CAMERA

00:21linkaihero.dev/skills documentation site ↗

10:57productAI Coding Crash Course (waitlist) ↗

00:00linkAI Hero Discord ↗

00:00channelMatt Pocock on Twitter/X ↗

FROM THE DESCRIPTION

PRIMARY CTAWhere the creator wants you to go next.

OTHER LINKSAlso linked in the description.

Storyboard