Your brain works differently. Your tools should too.
Hyperfocus restructures Claude's output so your brain can lock in. Not fewer words — better structure.
claude plugin marketplace add nextor2k/hyperfocus
claude plugin install hyperfocus@hyperfocusThen type /hyperfocus in any conversation. Done.
Hyperfocus changes how Claude talks to you.
Every response gets:
- Answer first — key point up top, context after
- Clear sections — subheadings you can scan in 2 seconds
- Short chunks — one idea per paragraph, short sentences
- Lists over walls — bullet points instead of dense prose
Important
Same depth. Same technical accuracy. Just structured so your brain can absorb it.
Pick your level of structure. Switch anytime with /hyperfocus clean|flow|zen.
For: code reviews, quick answers, PR descriptions.
Shorter paragraphs, front-loaded points, bullet lists. Professional tone stays intact.
For: learning, debugging, technical explanations.
Each section follows What → Why → How. Subheadings every 2–3 paragraphs for easy re-entry after breaks.
For: dense docs, architecture discussions, long sessions.
TL;DR at the top of every response. Almost everything in lists or tables. Every section stands alone.
This is what the same Claude response looks like without and with hyperfocus.
Without hyperfocus (default Claude):
The
useMemohook in React is used to memoize expensive computations so that they are only recalculated when their dependencies change. This is particularly useful when you have a component that re-renders frequently but has some computation that doesn't need to run on every render. WithoutuseMemo, React would recalculate the value on every render, which could lead to performance issues. You should use it when you have computationally expensive operations that depend on specific props or state values, but be careful not to overuse it as the memoization itself has a small cost.
With hyperfocus — all three modes side by side:
| Clean | Flow | Zen |
|---|---|---|
|
Use it when the computation is expensive and the component re-renders frequently. Don't overuse it — memoization itself has a small cost. |
What It Does It caches a computed value. React only recalculates when dependencies change. When To Use It Use
Watch Out Don't overuse it. Memoization has a small cost. Only apply it where you can measure a difference. |
TL;DR: What
When
Caution
|
Claude Code marketplace (recommended):
claude plugin marketplace add nextor2k/hyperfocus
claude plugin install hyperfocus@hyperfocusOr install the skill directly:
npx skills add nextor2k/hyperfocusOther agents (Codex, GitHub Copilot, manual)
Codex:
npx skills add nextor2k/hyperfocus -a codexGitHub Copilot:
npx skills add nextor2k/hyperfocus -a github-copilotManual: Clone this repo. Copy skills/hyperfocus/ into your project's skills folder.
| Action | Command |
|---|---|
| Activate | /hyperfocus |
| Switch mode | /hyperfocus clean or flow or zen |
| Turn off | stop hyperfocus |
| Always on | /hyperfocus persistent |
| Stop always on | /hyperfocus disable |
You can also say "focus mode", "adhd mode", or "adhd friendly" in your prompt.
Want hyperfocus on every session without typing /hyperfocus each time? Make it persistent.
/hyperfocus persistent
This installs a UserPromptSubmit hook that re-injects the formatting rules on every turn. Zero drift — even in long conversations.
| Command | What it does |
|---|---|
/hyperfocus persistent |
Auto-activate globally (all projects) |
/hyperfocus persistent project |
Auto-activate for this project only |
/hyperfocus disable |
Remove auto-activation |
Hyperfocus is a skill — one file of formatting rules that load into Claude's context. No code, no build step. Code blocks, commits, and PRs stay untouched. Only prose gets restructured.
- Claude Code — fully tested, mode persists all session
- Codex — supported, mode persists all session
- GitHub Copilot — supported via npx skills
Warning
Some agents (Opencode, omp) may need re-activation each turn. That's an agent limitation, not a hyperfocus one.
Every rule in Hyperfocus comes from peer-reviewed research or published accessibility standards.
| Principle | What the research says | Source |
|---|---|---|
| Short sentences | Comprehension drops sharply past 20 words | Oxford/Plain English |
| Chunking | ADHD brains process sequential chunks better than parallel info | PMC6996017 |
| Front-loading | ADHD readers may not reach the end — answer goes first | W3C COGA |
| Visual hierarchy | Good subheadings enable re-entry in under 2 seconds | ATG Publishing |
| Blank lines | Walls of text trigger task abandonment | ATG Level 1 |
Full source list
- W3C COGA — Making Content Usable for People with Cognitive and Learning Disabilities (2021)
- ATG Publishing — ADHD Friendly Publishing Standards
- WCAG 2.2 — Cognitive Accessibility (2023)
- MDN — Creating Effective Technical Documentation
- Readability Guidelines — Simple Sentences (Oxford Guide to Plain English)
- PMC — Constraints on Information Processing Capacity in Adults with ADHD (2020)
ADHD isn't a deficit to work around. Hyperfocus is the superpower.
This plugin removes the friction that blocks it. When information is structured right, your brain locks in — not because it was forced to, but because nothing got in the way.
If you have ideas for better ADHD-friendly formatting — especially if you have ADHD yourself — open an issue or PR.
