Big Idea

The argument in one line.

Claude Code's built-in Output Styles feature, not a different model, is the practical fix for Opus 5's increasingly jargon-dense default responses, and it can be swapped mid-conversation or set to persist per project.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You use Claude Code or Opus 5 daily and find yourself re-reading responses just to figure out what the model actually did.
You manage multiple codebases and want different explanation depths for a repo you know well versus one you're new to.
You tried Output Styles once when it launched, dismissed it as minor, and haven't touched it since.

SKIP IF…

You don't use Claude Code or Opus 5 — this is a tool-specific configuration walkthrough, not a general AI-writing tip.
You already switch output styles per project and per task as a matter of habit.

TL;DR

The full version, fast.

Opus 5's default responses have gotten more verbose and jargon-dense, to the point that even experienced engineers need to look up terms just to parse a routine update. The fix isn't a different model — it's Claude Code's underused Output Styles feature, which swaps the system prompt controlling explanation depth. The walkthrough covers installing a custom ELI5 style by tagging the Claude Code guide, switching styles with /config, and touring built-ins like Explanatory, which Anthropic's own team uses internally when onboarding engineers to a new repo. For text that's still too dense, STE100 (Simplified Technical English) trims it further. Styles persist per project, so different repos can default to different reading levels.

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 – 00:26

01 · Intro

Cold open on a genuinely confusing Opus 5 output; sets up the problem before naming the fix.

00:26 – 00:50

02 · The Blog Post

Reads from gruhn.me's 'Don't Be a Meat Proxy,' arguing AI output is extra effort to read: verbose, plausible-sounding nonsense, increasingly jargon-dense.

00:50 – 01:14

03 · Output Styles

Names the fix: Claude Code's Output Styles feature, covered once ~8-9 months earlier and mostly forgotten until now.

01:14 – 02:24

04 · Setup

Installs the ELI5 style by tagging the Claude Code guide and pasting the style text, then switches to it via /config.

02:24 – 03:09

05 · Built-In Styles

Tours Learning (makes you write code yourself) and Explanatory (what Anthropic tells new engineers to enable on an unfamiliar repo), illustrated with a Boris Cherny fireside-chat clip.

03:09 – 03:49

06 · Find Your Style

Uses /branch to fork the conversation and regenerate the same confusing output in five different styles side by side to find the best fit.

03:49 – 04:37

07 · STE100

Introduces Simplified Technical English, an existing industrial writing standard, as a stricter alternative when ELI5 still isn't plain enough.

04:37 – 05:35

08 · Per Project

Shows output styles persist per project via .claude/settings.local.json, so different repos keep different defaults simultaneously.

05:35 – 05:56

09 · Outro

Newsletter and Agentic Coding School plug, framed as staying ahead of models that keep getting more jargon-heavy by default.

Atomic Insights

Lines worth screenshotting.

Opus 5's default output has gotten measurably more jargon-dense, forcing readers to look up individual words just to parse routine status updates.
Claude Code's Output Styles feature rewrites the system prompt controlling how verbose or technical a response reads, without changing the underlying model.
Anthropic's own Claude Code team tells new hires to enable the Explanatory output style when they first touch an unfamiliar repo.
You can install a new output style by tagging the Claude Code guide and pasting the raw style text in — no manual file editing required.
/config switches output styles mid-session, and rewinding to an earlier prompt lets you replay it under a different style to compare readability directly.
STE100, or Simplified Technical English, is an existing industrial-writing standard for stripping jargon — repurposed here as a Claude Code output style, not something built for AI.
Output styles persist per project via .claude/settings.local.json, so one repo and another can default to completely different explanation depths at the same time.
/branch forks a conversation mid-stream and can regenerate the same confusing output in five different styles side by side, turning style selection into a direct comparison.
There's no single best output style — engineers reportedly switch between them depending on the project, the task, or how tired they are that day.

Takeaway

Pick an explanation depth, not just a model

READING AI OUTPUT

AI responses are getting more jargon-dense by default, and the practical fix is choosing how you want things explained to you, not just hoping the model writes more plainly.

02The Blog Post

AI output is getting harder to read by default, not because you're less sharp — verbosity and jargon density are measurably increasing across model updates.
If you have to look up individual words to parse a routine explanation, the problem is the output's calibration, not your comprehension.

03Output Styles

There's already a built-in fix for confusing AI output: a setting that controls explanation depth independent of which model you're running.
A feature you dismissed as minor when it launched can become essential months later once the underlying tool's default behavior shifts.

04Setup

You can install a new communication style for an AI assistant in under a minute by handing it the style text directly and asking it to apply it, with no manual config editing required.

05Built-In Styles

Teams that build AI coding tools use a more detailed explanation mode internally when onboarding people to unfamiliar code, a sign that better explanations, not less detail, is the actual fix for confusion.
There's a mode designed to have you write code yourself rather than just receive it, a reminder that the fix for jargon isn't always more clarity; sometimes it's more friction on purpose.

06Find Your Style

You can ask an AI to regenerate the same confusing answer in several different communication styles side by side, turning 'which style do I like' into a direct comparison instead of guesswork.

07STE100

Simplified Technical English is a decades-old industrial writing standard for stripping jargon from documentation, proof that old solutions to 'text is too dense' can be repurposed for AI output.

08Per Project

Explanation depth doesn't have to be a single global setting; different projects can default to different reading levels depending on how well you already know that codebase.

Glossary

Terms worth knowing.

Output Styles: A Claude Code feature that swaps the system prompt used to generate responses, changing how verbose, technical, or simplified the output reads without changing the underlying model.
ELI5: A custom output style (from 'explain like I'm 5') that forces short sentences, plain words, and minimal detail, meant for when the reader has little mental bandwidth left.
STE100: Simplified Technical English, an existing industrial writing standard that restricts vocabulary and sentence structure for clearer documentation, repurposed here as a Claude Code output style.
/config: The Claude Code slash command used to view and change settings, including which output style is currently active.
/branch: A Claude Code command that forks the current conversation into a new thread, letting you re-run the same prompt under different settings without losing the original.

Resources

Things they pointed at.

00:26linkDon't Be a Meat Proxy (blog post) ↗

00:50linkLydia Hallie's post on output styles ↗

02:40linkFireside Chat with Boris Cherny @Scale

03:49toolSTE100 / Simplified Technical English standard

Quotables

Lines you could clip.

00:20

“It feels like my English has become weaker or it's speaking another language.”

relatable, self-deprecating hook about AI output confusion→ TikTok hook↗ Tweet quote

00:30

“Reading AI output is extra effort. It's verbose, frequently contains all too plausible nonsense, and it is increasingly jargon dense.”

sharp, quotable critique lifted from the source blog post→ newsletter pull-quote↗ Tweet quote

02:40

“Whenever any new engineer joins our team, we tell them to use the exploratory output style.”

insider detail on how Anthropic's own Claude Code team onboards people→ IG reel cold open↗ 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.

metaphor

Okay. So one of the biggest problems that you've likely been facing with OPUS five recently is the fact it's really confusing when outputting any text. So for example, over here, I was chatting about some ideas, and it basically said the corpus is a engine.

Angles come from what you've already ingested. And I read for a lot of this output, and I'm kind of like, what on earth does this even mean? I have no idea.

It feels like my English has become weaker or it's speaking another language. And I know I'm not the only one feeling this way because there's a pretty recent viral blog post where basically the author says, reading AI output is extra effort. It's verbose, frequently contains all too plausible nonsense, and it is increasingly jargon dense.

And recently, they got this output from Claude, which I would not even try reading. And it said, had to look up every single word to make sense of this. And, basically, the solution that someone from the Claude Co.

Team recommended is using output styles. And I actually did make a video about this inside of my agent coding school almost, like, eight, nine months ago now back in October when it was released, but I haven't really used it since. So if I go over to output styles over here, I basically start out the video by saying that this is a feature that I don't use much, but it does exist.

But I have been using it more and more because this helps get around the fact that OPUS five is really annoying when it comes to outputting. So maybe quickly go through how we can set this up over here. So this will be down below, you can literally just copy the text.

But she essentially says that she likes to use this after a long day, honestly, helps so much. And it seems that some of the team members like using different output styles for either different projects, different tasks, or even at different times of the day depending on how tired they're feeling or how engaged in the process they want to be.

So it's not a case of you make one Apple style then you set it once. You will be switching back and forth between them over time. So, So anyways, I'll quickly go through how you can set this up.

Basically, this will be down below, so you can literally just copy the text. Then go over to claud code, tag the claud code guide by doing ads claud code guide, paste in the, uh, output style and say and say add this output style for me.

Press enter, and then it will go ahead and do that. And then once it is done, you can do slash config instead of cloud code. Search for output style over here, and then switch over to explain like m five, and then press escape.

And then you can just rewind, uh, to go back earlier one message, and then basically enter in the same prompt again. And then after entering the same prompt again like the explain like m five output style, it made much more sense.

Now you will have noticed that all other output styles as well. So there is a learning output style where it will ask you to write some code yourself. There is also the explanatory output style, which I know the Cloud Code team likes to use because they mentioned it in an interview where they get new hires to basically enable that when they're working inside of a brand new repo.

Whenever any new engineer joins our team, we tell them to use the exploratory output style.

So this is just like slash config output style equals exploratory. You run this in Quad Code, or you can ask Quad to set this for you.

And what it does is anytime Quad will make a change, it'll explain to you, hey, here's how the architecture works. Here's how this language works if you haven't used it before. Here's how this part of the code base works.

It'll explain to you so that you can learn. Now, of course, this is not a one size fits all Apple style. You may wanna kind of experiment around and find a better one for your liking.

And one way that you can do this is that anytime you come across some really confusing or annoying output, kind of like I have in this chat over here, you can basically do slash branch, and then it will branch from the existing conversation onto brand new chats. And then I can basically paste in the existing output style and say, hey.

So can you basically generate me, like, five of the output styles and how your previous output would look like in that brand new style? Essentially, goal here is to find a style of communication, which I understand.

And I can see there's a bunch of Apple styles, so there is a kid mode, which I will not be using. That's way too basic. There's also the Slack DM over here, which looks a bit better.

But, honestly, when I'm reading for it, then it's, like, still pretty confusing. And if you find that doesn't work for you, then you can ask it to make a APO style based on ASD s t e 100, which is a language standard, which is for simplified technical English.

And if I were to basically rewind and then say, like, okay. Can you explain explain in this instead?

Then it will give a much better style. And you may wanna adjust this slightly. And if I were to make this into a brand new Apple style, then I may find over time, actually, I want it to be a bit more technical.

And now reading through this, I can be like, okay. Wow. This is significantly better.

I wanna turn this into Apple style. Tag the cloud code guide and say, this into a Apple style.

And I may notice over time, like, hey. I'm getting more tentacle or something like that, in which case I may wanna bump up the level of the output style in some way. Now one of the nice built in features into Cloud Code is the fact that the output styles persist per project.

So you can see for agent stack over here, when I go to settings dot local dot JSON inside the dot Cloud folder, then I have output style set to explain like m five. But if I now set the output style for this other project to the s t 100, uh, in simple technical English, then I can see that the previous project's output style has not changed, but this new one over here has changed to s t 100.

So this is really good because you can have different output styles for different projects. Anyways, for me personally, I will be using it for different projects, different output styles. So for example, recently, I have been working on a new project of building really good agent sandboxes.

So I can do slash config and send that to explanatory to understand more about agent sandboxes. But for projects I'm very familiar with, I can have it be way shorter instead. Anyways, I would basically recommend playing around because I think this is something we will need to do more going forward as the models are getting slightly more confusing in their default outputs.

Anyways, if you do like this kind of stuff, then do subscribe to the channel. And if you do wanna get more tips from me on a regular basis, then do check out my email newsletter. It will be linked down below.

By signing up, you will also get access to a bunch of free videos from my agentic coding school class as well. And you're also welcome to reply to any of the issues, and I will try to get back to you as well because I really like hearing from people replying to newsletters.

The Hook

The bait, then the rug-pull.

Ray Amjad opens with an Opus 5 output he can barely parse himself, then spends five minutes on the fix Anthropic's own team uses internally: Claude Code's Output Styles.

Frameworks

Named ideas worth stealing.

03:49list

The Five-Style Comparison

Kid Mode (ELI5)
The Slack DM
Explanatory
Learning
STE100

Five ways to have Claude re-explain the same confusing output, generated live via /branch so they can be compared side by side instead of guessed at.

Steal forany AI tool output you find hard to parse

CTA Breakdown

How they asked for the click.

VERBAL ASK

05:35newsletter

“if you do wanna get more tips from me on a regular basis, then do check out my email newsletter... by signing up, you will also get access to a bunch of free videos from my agentic coding school class”

soft close at the very end — offers free bonus content as the incentive to subscribe rather than a hard sell, paired with a subscribe ask

MENTIONED ON CAMERA

00:26linkDon't Be a Meat Proxy (blog post) ↗

00:50linkLydia Hallie's post on output styles ↗

FROM THE DESCRIPTION

PRIMARY CTAWhere the creator wants you to go next.