Coding agents tend to bury the answer: a compliment for the question, a paragraph of context, a tangent about your dependencies, then the fix, then an offer to dig deeper. i-have-adhd rewrites that habit. With it on, your agent leads with the next action, numbers multi-step work, ends with one concrete next step and drops the preamble and recaps. The README's before and after shows a rambling paragraph becoming one line plus three numbered steps.
It is ten rules in one skill: lead with the next action, number steps, end with a next step, suppress tangents, restate where you are each turn, give specific time estimates, make wins visible, report errors matter-of-factly, cap lists at five items, and skip preamble and closers. It also says to explain fully when you ask, confirm before destructive actions, and stop after three failed fixes. The rules are loosely based on the book The Adult ADHD Tool Kit, but you do not need ADHD to like shorter answers.
It works with Claude Code, Codex, Gemini CLI, GitHub Copilot, Antigravity and other agents. It is MIT-licensed, with about 61,000 stars and 53 contributors since May 2026, and version 0.4.1 came out on 19 October 2026.
- Repository: github.com/ayghri/i-have-adhd
- Licence: MIT (MIT License)
- Language: Python. Stars: 61.1K. Forks: 3,578. Last push: Oct 19, 2026.
- Scan: safe, Nov 13, 2026, commit 1ecea19
Who it is for
Anyone who skims long AI answers looking for the command to run, people with ADHD or attention fatigue, and developers who want terse, actionable replies from Claude Code or another agent.
Getting started
1. In a terminal, add the plugin's marketplace to Claude Code
claude plugin marketplace add ayghri/i-have-adhd2. Install it
claude plugin install i-have-adhd@i-have-adhd3. Turn it on in a session
/i-have-adhd4. Optional: make it always on in Claude Code
touch ~/.claude/.i-have-adhd-alwaysSay stop adhd mode or normal mode to switch it off for a session, and delete the flag file to turn always-on off. INSTALL.md covers Codex, Gemini CLI, Copilot and other agents.
Safety scan
We cloned ayghri/i-have-adhd at commit 1ecea19 on Nov 13, 2026 and ran the checks described on the GitHub Tools page: credential patterns, decode-and-execute code, install-time scripts, committed binaries, risky CI workflows, every host the code talks to, known vulnerabilities in pinned dependencies, and project hygiene. A person read every hit. This is what we found.
- No secret hits, no pattern hits, no binaries and no lockfiles across 74 files and about 13,500 lines, mostly Markdown, with small TypeScript, JavaScript and Python helpers.
- The Claude Code plugin has one SessionStart hook. It checks whether ~/.claude/.i-have-adhd-always exists; if not, it exits. If you created that file, it prints the skill's rules into the session. That is all it does: it reads one file from the plugin folder and touches the network not at all. Shell and PowerShell copies do the same for systems without Node.
- The extra pass for skill repos was clean. The hidden characters are zero-width non-joiners that Persian spelling needs, in the translated install guides. No injection phrases, exfiltration hosts, credential paths or pipe-to-shell. The scanner found no network hosts in the code at all.
- The scripts and evals folders hold the maintainers' checks and evaluation harness for testing the rules against models; installing the skill does not run them.
- Four workflows, no pull_request_target. The one third-party action is pinned to a tag rather than a commit. A licence and contributing guide are present; no security policy.
What the scanner counted
| Check | Result |
|---|---|
| Secrets | None found. |
| Suspicious code | None found. |
| Install-time code | None: nothing runs at install beyond the package manager itself. |
| Committed binaries | None. |
| CI workflows | 4 workflows. None use pull_request_target. 1 of 1 third-party action pinned to a tag rather than a commit. |
| Network hosts | 0 distinct hosts referenced from source. No URLs to bare IP addresses. |
| Known vulnerabilities | No lockfile to check: dependencies are declared as ranges, so what gets installed is whatever is current on the day. |
| Project hygiene | Has licence file, contributing guide. Missing security policy, automated dependency updates, CodeQL. |
| OpenSSF Scorecard | Not scored: the project is not in Scorecard's weekly index. |
By the numbers
| Stars | 61.1K |
|---|---|
| Forks | 3,578 |
| Contributors | 53 |
| Commits | 263 |
| Open issues | 33 |
| Open pull requests | 66 |
| Releases | 1 |
| Latest release | v0.4.1 |
| Licence | MIT |
| Main language | Python |
| Project age | 6 months |
| Last push | Oct 19, 2026 |
| Tracked files | 74 |
| Lines of code | 13.5K |
| Checkout size | 623 KB |
Lines by language: Markdown 9,276, Python 3,127, TypeScript 391, JavaScript 265, JSON 190, YAML 157.
Questions
Is i-have-adhd free?
Yes. It is free and MIT-licensed, with no account.
Do I need ADHD to use it?
No. The README says no diagnosis is needed. It suits anyone who wants answers that start with what to do.
Will my agent still explain things when I ask?
Yes. The rules make an exception for explanations you request, and they keep the agent asking before destructive actions.
Is this a medical or therapy tool?
No. It only changes how your AI agent formats replies. It draws on ADHD coping strategies from a published book, adapted to how an assistant should write.
This post is part of GitHub Tools, where every repository is cloned and scanned before it is written up. The scan is a snapshot of one commit on one day; the repository has moved on since, so check it before you install.