i-have-adhd

i-have-adhd is a skill (also packaged as a Claude Code plugin) that changes how a coding assistant formats its replies. Instead of opening with "Great question!" and burying the fix three paragraphs down, the assistant leads with the action, numbers the steps, and stops talking once the task is done. It's built for developers who use AI coding assistants daily and find the verbose, hedging style of typical LLM output slows them down.

The project isn't a code library or API. It's a prompt/skill definition (SKILL.md) plus install instructions that tell a coding agent to follow a specific set of response rules. It works with Claude Code's plugin system, and the AGENTS.md file documents how to wire it into other agent setups.

Key features

• Action-first responses: the assistant states the command or edit to make before explaining anything.
• Numbered steps: multi-step tasks get an explicit numbered list instead of a wall of prose.
• One concrete next step: every reply ends with a single, specific thing to do next, not a vague "let me know if you need more."
• Tangent suppression: side notes and unrelated observations get cut.
• State tracking: the assistant restates current state each turn so context isn't lost.
• Specific estimates: time estimates are given in minutes, not "a bit" or "soon."
• Capped lists: lists are limited to 5 items to avoid scroll fatigue.
• No preamble or closers: no "Hope this helps!", no recap, no filler openers.
• Matter-of-fact error handling: errors are reported plainly, without softening language.
• Editable rule set: the 10 rules live in a single SKILL.md file you can fork and tune.

Ideal use cases

This fits developers who use Claude Code (or a compatible coding agent) and are tired of re-reading responses to find the actual instruction. It's useful for anyone who processes information better with direct, structured output, including people with ADHD, but the skill itself works for anyone who wants less scrolling and more doing.

It's a good fit for fast iteration loops: debugging, running migrations, applying config changes, anything where you want the command first and the explanation (if any) after. It's less useful if you rely on your assistant's verbose reasoning to learn why a fix works, since the rules explicitly suppress tangents and extended explanation. If you want a chatty, exploratory assistant that walks through trade-offs before acting, this skill will feel abrupt.

It also isn't a standalone tool or service. It only changes behavior when installed as a skill/plugin inside a compatible coding agent (Claude Code is the primary target). If you don't use such an agent, there's nothing here to run.

Installation

The simplest install is a copy/paste prompt into your CLI-based coding agent:

Install the i-have-adhd skill/plugin from https://github.com/ayghri/i-have-adhd, refer to the repo's AGENTS.md for instructions.

Alternatively, follow the steps in INSTALL.md in the repo for a manual setup.

To customize the rules, fork the repo, edit skills/i-have-adhd/SKILL.md, then swap in your fork inside Claude Code:

claude plugin uninstall i-have-adhd            # drop the upstream copy first
claude plugin marketplace remove i-have-adhd   # fork and upstream share both names
claude plugin marketplace add <your-username>/i-have-adhd
claude plugin install i-have-adhd@i-have-adhd

Restart Claude Code after installing, then re-invoke /i-have-adhd to activate the skill in your session.