How to Change Claude Code’s Output Style With a Simple ELI5 Prompt

I am not a programmer. So when Claude Code finishes a job and gives me a wall of coding language, I can lose the plot pretty quickly.

I do not need a victory speech. I need three things: what did you do, did it work, and what do I need to do next?

The work is done. The explanation is the problem.

Claude Code can be very good at the actual work and still give you a final answer that is longer or more technical than you need. That is especially unhelpful when you are learning, tired, or trying to make a quick decision without pretending you understand every coding term on the screen.

I kept asking it to use plain English. Then I learned I could save that preference as an output style instead of repeating myself in every conversation.

What is a Claude Code output style?

An output style changes how Claude Code responds. It can change the tone, format, and amount of explanation without changing what Claude knows. Claude Code includes a few built-in styles, and it also lets you create your own.

For this one, I want Claude Code to keep doing the coding work. I just want it to explain the result like a normal person. That is why the style keeps the built-in coding instructions and changes the way the answer is written.

You can read the technical explanation in the official Claude Code output styles documentation ↗.

Copy and paste this into Claude Code

Paste the full prompt below into Claude Code while you are inside a project. It tells Claude where to save the custom style, turns it on globally, turns it on for the current project, and asks it to leave the rest of your settings alone.

Set my Claude Code Output Style to the one below.
Save it as ELI5.md in my global ~/.claude/output-styles/ folder.
Then set "outputStyle" to "ELI5" in my global ~/.claude/settings.json file and in this project's .claude/settings.local.json file.
Create the folder, file, or key if it is missing.
Do not overwrite or break any existing JSON.
Then list the files you changed and tell me to restart Claude Code.

---
name: ELI5
description: Keep it simple, please
keep-coding-instructions: true
---

It has been a long day and my brain is fried. Talk to me like I am 5.

Only report to me in ASD-STE100 Simplified Technical English. Use approved plain words, active voice, and one instruction or idea per sentence.

Use small words, short sentences, and short paragraphs. If you have to use a big word, explain it right after. Only return what is actually necessary.

Just tell me what you did, whether it worked, and what I should do now.

If I have to decide something, give me two options at most. Give me the context I need to pick fast. Tell me which option you recommend.

Keep paths and commands exact. I have no brain cells left for the rest.

After Claude lists the files it changed, restart Claude Code or use /clear. Output styles load when a new session starts, so the change may not show up in the conversation that installed it.

What this changes

• Shorter answers. Claude is told to leave out the parts you do not need.
• Plain language. It uses shorter words and explains a technical term when the term is necessary.
• A clear status. It tells you what changed and whether it worked.
• A useful next step. It tells you what you should do now.
• Fewer decision dumps. If you need to choose, it gives you two options at most and tells you which one it recommends.
• Exact commands. The explanation gets simpler, but file paths and commands stay exact.

The style is global, so Claude Code can find it in your other projects. A project can still have its own setting that overrides the global choice.

Will a simpler output style use fewer tokens?

It can use fewer output tokens because you are asking Claude to write less. It can also feel faster because there is less final text to stream onto the screen.

But I would not promise that every task becomes cheaper or faster. The instructions in the output style add a little context, and Claude still has to inspect files, use tools, write code, and do whatever the job requires. This changes the report you get at the end. It does not magically make the work disappear.

A few quick questions