i-have-adhd: a Claude Code output style that leads with the next action

i-have-adhd is an output-style skill by ayghri, with 14,235 installs in the Skillselion catalog, that reshapes every Claude Code reply to be action-first, numbered and free of preamble. The catalog rates it quality tier A and LOW risk on 2026-09-13, and the GitHub repo carries 43,673 stars under an MIT license.

Claude Code gives you the right answer and you still cannot act on it, because the one command you needed sits in paragraph three behind "Great question! Let me think about this." The i-have-adhd skill by ayghri rewrites the shape of every reply instead of its content, and it leads every ADHD-related skill in the Docs & Planning category with 14,235 installs.

• i-have-adhd records 14,235 installs, quality tier A, LOW risk and stage grow on its Skillselion listing (Skillselion catalog API, 2026-09-13).
• The repository carries 43,673 stars, 2,494 forks and an MIT license (GitHub API, 2026-09-13). The catalog snapshot of the same repo reads 27,229 stars, so treat the catalog figure as a lagging mirror of GitHub, and note that star counts are repo-wide rather than per-skill.
• The skill is 10 numbered rules plus a 5-item pre-send deletion check, shipped as version 0.3.0 in the repo package.json.
• Its frontmatter sets disable-model-invocation: true. Claude will never reach for it on its own, so you type /i-have-adhd or nothing changes.
• Twelve skill listings match the query "adhd" in a catalog of 60,442 skills and 82,914 total listings, and i-have-adhd holds 2.5 times the installs of the runner-up (Skillselion catalog API, 2026-09-13).

i-have-adhd is an output-style skill: it changes how Claude writes every response, not what Claude knows. The catalog files it under Productivity & Planning with a LOW risk rating, quality tier A and a repo last pushed on 2026-09-10 (GitHub API, 2026-09-13), which puts it among the actively maintained entries in that bucket. It adds no tools, no MCP server and no framework knowledge. The whole payload is one SKILL.md of formatting doctrine, which is also why it costs almost nothing to try.

The author frames the rules around five stated facts about ADHD reading, including that working memory is small so anything off screen is forgotten, and that knowing the answer is not the same as doing the answer.

i-have-adhd enforces ten numbered rules, from leading with the next concrete action to cutting every closing pleasantry. In the order they appear in the published SKILL.md:

Lead with the next action. The first line is a command, path or snippet, not context.

Number multi-step tasks, one bounded action per step.

End with one concrete next action doable in under two minutes.

Suppress tangents. Finish the first issue, then offer the second as a separate question.

Restate state every turn, in the form "Step 3 of 5 done".

Give specific time estimates in concrete units, never "a bit of work".

Make completed work visible: show what now runs, with the command to prove it.

Keep error reports matter of fact. No "Uh oh", just location, cause and fix.

Cap visible lists to 5 items, while keeping the rest internally rather than discarding them.

No preamble, no recap, no closing pleasantries.

Then a pre-send check deletes five things before the reply leaves: an opening sentence that announces what is about to happen, a closing "anything else?", any "by the way" sidebar, hedging adverbs that carry no information, and any idiom such as "circle back". The final test is whether the first line and the last line alone tell you what to do next and what just happened.

i-have-adhd by ayghri, with 14,235 installs (Skillselion catalog API, 2026-09-13) and 43,673 GitHub stars (GitHub API, 2026-09-13). Install it with the two commands from the repo INSTALL.md:

claude plugin marketplace add ayghri/i-have-adhd then claude plugin install i-have-adhd@i-have-adhd

Two alternatives are worth knowing. adhd by uditakhourii has 5,726 installs and 4,059 stars, but it does a different job: it pushes the agent toward divergent, non-obvious solutions rather than reshaping output. adhd-daily-planner by curiositech has 267 installs and 211 stars and plans a human day, not an agent reply. For the wider bucket, the ranked list lives on the best skills for Docs & Planning.

ayghri, author of the i-have-adhd skill for Claude Code

Type /i-have-adhd once and the rules persist for the rest of the session, until you say "stop adhd mode" or "normal mode". Because the frontmatter carries disable-model-invocation: true, nothing happens until you invoke it, which matches how Claude Code treats skills that opt out of model invocation (Claude Code skills documentation). If you want it always on, the repo ships a SessionStart hook that loads the ruleset at the start of every session once you run touch ~/.claude/.i-have-adhd-always.

Any developer who reads agent output under time pressure, which is most of them. The repo tagline says no ADHD diagnosis is needed, and the rules map cleanly onto ordinary code-review fatigue: action first, steps numbered, state restated. The skill also names six conditions where it stands down, including an explicit request to explain, a destructive action such as a schema migration, a debug spiral after three failed turns, and a rule that would delete the answer itself. That last exception matters. Asked "what are my options", the skill returns 2 to 4 ranked options rather than one path.

i-have-adhd (14,235 installs) for the reply shape itself.

planning-with-files by OthmanAdi, 42,710 installs, for file-based task plans that survive a session reset. Rule 5 restates state; this one stores it.

planning-and-task-breakdown by addyosmani, 33,435 installs, to turn a spec into the ordered, verifiable steps rule 2 wants to number.

writing-clearly-and-concisely by softaworks, 4,147 installs, for the docs and commit messages the output style never touches.

All four install figures come from the Skillselion catalog API on 2026-09-13.

Installing it and expecting a change. With disable-model-invocation: true, the skill is inert until you type /i-have-adhd in the session.

Running two output-style skills at once. They issue contradictory formatting rules and the model picks arbitrarily. Pick one and keep the others for their tools.

Reading catalog star counts as a quality score for one skill. The 27,229 figure on the listing is repo-wide, and a monorepo with 40 skills inherits the same number on all of them. Check installs, which are per skill, and confirm against GitHub.

Type /i-have-adhd once, read the next reply's first and last line, and you will know in one turn whether this belongs in your setup. If you would rather write the rules yourself, start from how to build custom Claude Code skills.

How do I install the i-have-adhd skill in Claude Code?

Run claude plugin marketplace add ayghri/i-have-adhd and then claude plugin install i-have-adhd@i-have-adhd, the two commands the repo INSTALL.md gives for Claude Code. The skill is MIT licensed and rated LOW risk with 14,235 installs in the catalog on 2026-09-13. Full details on the i-have-adhd listing.

Why does the i-have-adhd skill do nothing after I install it?

Its frontmatter sets disable-model-invocation: true, so Claude never applies it automatically. Type /i-have-adhd in the session, or run touch ~/.claude/.i-have-adhd-always to load the ruleset from the repo SessionStart hook every session. Other skills that opt out of model invocation behave the same way, so check the listing before you assume a skill is broken: see the Docs and Planning category.

How many rules does the i-have-adhd skill have?

Ten, plus a five-item pre-send deletion check that strips announcing openers, closing pleasantries, sidebars, empty hedges and idioms. Version 0.3.0 of the repo publishes all of them in one SKILL.md, which is why the skill adds no tools and no dependencies. Compare it with the rest of the best skills for Docs & Planning.

Is i-have-adhd the most popular ADHD skill for Claude Code?

Yes, by a wide margin. It holds 14,235 installs against 5,726 for the runner-up among the twelve skill listings matching "adhd" in the catalog on 2026-09-13. The two skills solve different problems, so read the runner-up too: adhd by uditakhourii.

Does i-have-adhd change what Claude knows or only how it writes?

Only how it writes. It is an output-style skill with no tools, no MCP server and no framework knowledge, so it applies equally to a refactor and to casual chat. If you want a skill that also carries procedure, see how to build custom Claude Code skills.

By Skillselion, an Ellelion LLC publication · Updated September 13, 2026 · 5 min read · Stats verified against the live catalog.

Ranked by Skillselion - an independent directory of AI-coding tools, not affiliated with Anthropic, OpenAI or Cursor. Tool rankings reflect real adoption (installs, then GitHub stars) from the skills.sh registry and GitHub, last updated September 13, 2026.