Claude Has an ADHD Mode, and It's Genius
The free tool with 18,000 GitHub stars that stops Claude from burying the answer. Two real ways to get it, whether or not you use Claude Code, and why it helps far beyond ADHD.
Save it, print it, read it offline. It's yours.
Most AI answers bury the point. You ask a simple question and get three paragraphs of “great question, let’s explore this” before you ever reach anything useful. A developer got sick of that, built a fix, and put it on GitHub for free. It’s called i-have-adhd, and it’s sitting at over 18,000 stars.
You don't need an ADHD diagnosis to use this or benefit from it. The rules underneath it are what a genuinely clear answer looks like, direct, numbered, no filler. Most people want this, it took an ADHD-focused developer to build it. The project says it better than I can:
The skill’s own description says it best: output isn’t only shorter, it’s shaped so it’s easy to act on. Once it’s active, every response follows 10 rules:
- Lead with actionable steps, not context
- Number multi-step tasks, one bounded action per step
- End with one concrete next action that takes under 2 minutes
- No tangents, finish the current thing first
- Restate progress and the current step in each message
- Give specific time estimates (“15 minutes,” not “a while”)
- Make completed work visible and concrete
- State errors matter-of-factly, no “uh oh”
- Cap any list at 5 items
- No preamble, no recap, no closing pleasantries
It’s not rigid about it either. The rules step aside for real explanations, anything destructive, debugging spirals, or genuine ambiguity, it’s built to be direct, not robotic.
You ask: "How should I follow up with a client who hasn't responded in a week?"
Before: "Great question! Following up with clients can feel tricky, you want to stay polite while making sure your message doesn't get lost. There are a few different approaches depending on the relationship and how much time has passed, so let's think through this..." (still hasn't given you anything to send)
After: "Send this: 'Hi [Name], floating this back to the top of your inbox in case it got buried, happy to answer any questions.' That's it."
It’s a Claude Code plugin, which means you need Claude Code, not regular Claude.ai chat, to install it this way.
Getting Claude Code itself, if you don’t have it:
- Go to claude.com/download and download the desktop app for your computer, Mac, Windows, and Linux are all supported
- Install it like any other app (double-click the file, follow the prompts)
- Sign in with your Claude account
One real thing to know before you start: Claude Code needs a Claude Pro subscription ($20/month) or API credits. Regular Claude chat is free, but Claude Code specifically isn't. Know that going in so you're not surprised partway through.
Installing the skill:
Whether you’re using the desktop app or a terminal window, Claude Code works the same way, you type commands right into the chat box. Open Claude Code and type this:
/plugin marketplace add ayghri/i-have-adhd
Then type this:
/plugin install i-have-adhd@i-have-adhd
You’ll see a confirmation. If it doesn’t say “Plugin is now active,” type /reload-plugins and it will be.
Now turn it on:
/i-have-adhd
The rules apply for the rest of that session. Say “stop adhd mode” or “normal mode” any time to turn it off, or remove it completely, same way, typed right into the chat:
/plugin uninstall i-have-adhd@i-have-adhd
/plugin marketplace remove i-have-adhd
If you don’t use Claude Code and don’t want to set it up, you can still get the same behavior in regular Claude. This isn’t the official GitHub tool itself, that one only runs inside Claude Code, but it’s the same 10 rules turned into a plain instruction you paste at the start of any chat.
From now on, format your answers this way: lead with the actionable step, not the context. Number any multi-step process, one action per step. End every answer with one concrete next action I can do in under 2 minutes. No tangents, finish the current point before raising a new one. Restate what's done and what's next whenever we're mid-task. Give specific time estimates, not vague ones. Make finished work visible and concrete, don't say it's done, show it. State any errors or problems plainly, don't soften them. Cap any list at 5 items. Cut the preamble, the recap, and the closing pleasantries. Give me the answer, nothing else.
Paste it once, then ask your next question.
The takeaway
One free, real tool, 18,000 people already voted with a star, and two honest ways to get the same result. Use the real Claude Code plugin if you’ve already got Claude Code, or the plain instruction above if you don’t. Either way, the difference shows up in the first response.
Want more like this?
One simple AI win in your inbox every Sunday. No jargon, no overwhelm, unsubscribe whenever you like.
Want more like this?
Free AI guides added regularly. No jargon, no overwhelm.
kellystrattonai.com
