i-have-adhd is an open-source skill that makes a coding agent's answer start with "what to do now" on the very first line. Ayoub Ghriss (@AyGhriTweets), a machine learning researcher at the University of Colorado Boulder, released it on May 14, 2026, and it had 52,051 GitHub stars as of September 29, 2026. Despite ADHD being in the name, the README says "ADHD-friendly outputs. No ADHD diagnosis needed!" Anyone can use it without a diagnosis. It states the next action first, numbers multi-step work, and trims long preambles, recaps, and sign-offs. In Claude Code, Codex, and others, you can get started by pasting a single install sentence from the repo.
How answers change when you turn on the i-have-adhd skill
The README compares answers to a request to fix an auth flow. Before opens with a long explanation of background and ends with a sign-off. After puts the command and file location first, then walks through the order of steps.
| Before | After |
|---|---|
"Good question. The auth flow involves middleware, token verification, and cookie handling. verifyToken in src/auth.ts seems to use an outdated API. Update the package, rewrite the function, and test it. Hope this helps!" | "Run npm install jsonwebtoken@latest, then edit src/auth.ts:42.1. Open src/auth.ts.2. Replace verifyToken (lines 42-58).3. Run npm test -- auth.spec.ts.Next: if the test fails, paste the first error line. |
The 10 rules
- Start the first line with the next action.
- Number multi-step work and put one action in each step.
- If work is unfinished, end with a next action you can do in 2 minutes.
- Handle side topics after finishing the current problem.
- For multi-turn work, restate progress every turn.
- Estimate time concretely, in minutes or hours.
- After a change, say what now works.
- For errors, state the location, cause, and fix in a calm tone.
- Keep visible lists to 5 items or fewer per group.
- Trim unnecessary preamble, recaps, and sign-offs.
If you ask for an explanation, it explains fully, it confirms before destructive actions, and if three fixes fail to solve it, it stops and states the assumption it doubts. If a request is ambiguous, it asks one short question. The README says it adapts the book The Adult ADHD Tool Kit by Russell Ramsay and Anthony Rostain to how LLMs respond.
Does accuracy hold up when answers get shorter
The repo's evals/ folder evaluated 14 cases, 3 runs each, on August 2, 2026 with claude-opus-4-8 and Claude Code 2.1.220. The same model handled generation and grading, and grading hid the conditions and shuffled the response order. All five category scores went up, and across the 14 cases it won 10, tied 2, and lost 2.
| Category | Weight | Without the skill | With the skill |
|---|---|---|---|
| Accuracy | 35% | 4.333 | 4.524 |
| Autonomy | 25% | 3.762 | 4.167 |
| Actionability | 20% | 3.905 | 4.619 |
| Safety | 10% | 4.643 | 4.667 |
| Brevity | 10% | 3.429 | 4.571 |
| Weighted total | 4.045 | 4.473 |
The cases with the biggest differences were multi-step progress reports (2.23→4.77) and error reports (2.07→4.47). The developer wrote in the results file, "we do not trade accuracy for brevity."
Install: paste one sentence
First, paste the following sentence into your agent so it installs following the repo's instructions.
Install the i-have-adhd skill/plugin from https://github.com/ayghri/i-have-adhd, refer to the repo's AGENTS.md for instructions.Claude Code
claude plugin marketplace add ayghri/i-have-adhd
claude plugin install i-have-adhd@i-have-adhd
claude plugin list
# Invoke: /i-have-adhd
# Update
claude plugin marketplace update i-have-adhd
# Remove
claude plugin uninstall i-have-adhd
claude plugin marketplace remove i-have-adhdCodex
codex plugin marketplace add ayghri/i-have-adhd --ref main
codex plugin add i-have-adhd@i-have-adhd
codex plugin list
# Invoke: $i-have-adhd
# Update
codex plugin marketplace upgrade i-have-adhd
codex plugin remove i-have-adhd
codex plugin add i-have-adhd@i-have-adhd
# Remove
codex plugin remove i-have-adhd
codex plugin marketplace remove i-have-adhdGemini CLI
mkdir -p ~/.gemini/commands
curl -fsSL https://raw.githubusercontent.com/ayghri/i-have-adhd/main/skills/i-have-adhd/agents/gemini.toml -o ~/.gemini/commands/i-have-adhd.toml
# Invoke: /i-have-adhd
ls ~/.gemini/commands
# Update: rerun the curl command above
# Remove
rm ~/.gemini/commands/i-have-adhd.tomlThe extension is always on:
gemini extensions install https://github.com/ayghri/i-have-adhd
gemini extensions list
gemini extensions update i-have-adhd
gemini extensions uninstall i-have-adhdCursor
npx skills add ayghri/i-have-adhd -a cursor -y
npx skills list
npx skills update i-have-adhd
npx skills remove i-have-adhd
# Invoke: /i-have-adhd in a new chatFor a full project install, add -g to the command.
| Tool | Install | Invoke |
|---|---|---|
| OpenCode | git clone https://github.com/ayghri/i-have-adhd ~/.config/opencode/vendor/i-have-adhd, then add the plugin path to the config file | /i-have-adhd |
| Antigravity | agy plugin install https://github.com/ayghri/i-have-adhd | Check with agy plugin list |
| GitHub Copilot | npx skills add ayghri/i-have-adhd -a github-copilot | /i-have-adhd |
| Grok | grok plugin install ayghri/i-have-adhd --trust, then enable | /i-have-adhd |
| Pi | pi install https://github.com/ayghri/i-have-adhd | /i-have-adhd |
| Oh My Pi | omp plugin marketplace add ayghri/i-have-adhd, then omp plugin install --scope user i-have-adhd@i-have-adhd | /i-have-adhd |
| Qwen Code | qwen extensions install ayghri/i-have-adhd | /i-have-adhd |
| Kimi Code CLI | /plugins → Custom → repo URL → Trust and install | /skill:i-have-adhd |
| Zed | Import the repo's SKILL.md URL in the Skills manager | /i-have-adhd |
| Hermes | hermes skills install ayghri/i-have-adhd/skills/i-have-adhd | /i-have-adhd |
| AstronClaw | Upload SKILL.md to My skills | Turn the skill on |
There are also a Korean README and a Korean install guide.
Keep it always on
By default it applies when you invoke it directly. Once you call it, it keeps applying in that conversation until you say "stop adhd mode" or "normal mode." In Claude Code, you turn it always on with a flag file and turn it off by deleting it.
touch ~/.claude/.i-have-adhd-always
# Turn off
rm ~/.claude/.i-have-adhd-alwaysFor Codex, put the "Output style" block from INSTALL.md into ~/.codex/AGENTS.md, and for other tools, into their own instruction files.
## Output style
The reader has ADHD. Shape every response so it can be acted on:
1. Lead with the answer or next action: command, path, or snippet first.
2. Number multi-step work; one bounded action per step.
3. End with one next action doable in under two minutes.
4. Finish the current issue before raising a new one.
5. Restate progress each turn ("step 3 of 5 done").
6. Give time estimates in concrete units, never "a bit".
7. After a change, show what now works.
8. Errors: state location, cause, and fix. No drama.
9. Cap lists to 5 items.
10. No preamble, no recaps, no closers.
Exceptions: explain fully when asked to explain. Confirm before destructive actions. After three failed fixes, stop and name the doubtful assumption. If the request is ambiguous, ask one short question.OpenCode supports the flag file ~/.config/opencode/.i-have-adhd-always, and Pi supports ~/.pi/agent/.i-have-adhd-always.
Adapt it to your own style
Fork the repo, edit skills/i-have-adhd/SKILL.md, then swap yours in as follows.
claude plugin uninstall i-have-adhd
claude plugin marketplace remove i-have-adhd
claude plugin marketplace add <your-username>/i-have-adhd
claude plugin install i-have-adhd@i-have-adhdAfter restarting, call /i-have-adhd again.
Watch the video
It is an approximately 19-minute video released on August 20, 2026 (KST) by Kacper Rutkiewicz of AI Made Simple, which compares two Claude Code sessions given the same question.
Related posts
- Two custom instructions for readable AI answers — how to use "talk to me like I have ADHD" as a one-line prompt
- Claude Code effort guide
- Coding agent