Overview
i-Have-ADHD is a community skill/plugin (16K+ GitHub stars) that restructures coding assistant responses to lead with concrete actions instead of verbose preamble. It works across Claude Code, Codex, and other coding agents, enforcing numbered step sequences, time estimates, and suppression of tangential commentary.
Features
- ✓Lead with next action
- ✓Number multi-step tasks
- ✓Specific time estimates
- ✓Suppress tangential commentary
- ✓Cap lists at 5 items
- ✓Works across multiple coding agents
Installation
claude plugin marketplace add ayghri/i-have-adhd && claude plugin install i-have-adhd@i-have-adhdPros
- +Eliminates verbose preamble in agent output
- +16K+ stars, highly active community
- +Works across Claude Code, Codex, and other agents
- +Easily forkable and customizable
- +MIT licensed, free to use
Cons
- −Formatting-only; does not improve reasoning
- −Style may not suit all users
- −Early-stage; installation paths may shift
Alternatives
Documentation
i-Have-ADHD
Overview
i-Have-ADHD is a community skill / plugin designed to restructure coding assistant responses so they are "ADHD-friendly": concise, action-led, and free of fluff. Instead of burying the answer under pages of preamble like "Great question!" or "Hope this helps!", the plugin enforces output that leads with the next concrete action and numbers multi-step sequences.
It has surged to over 16,000 GitHub stars and 893 forks, making it one of the most-trending developer productivity tools of early 2026. It works across multiple coding agents including Claude Code, Codex, and others, with each supporting the same formatting discipline.
The philosophy behind the project is simple: coding agents spend too many tokens summarizing what's already obvious, and the useful answer gets pushed to the bottom. i-Have-ADHD flips the priority — action first, explanation only if it adds value.
Features
- Lead with the next action: the first line of output is always the concrete next step
- Number multi-step tasks: every task with more than one step gets numbered instructions
- End with a concrete next step: no open-ended closers like "let me know if you have questions"
- Suppress tangents: tangential commentary is dropped unless directly relevant
- Restate state every turn: the agent always re-establishes current context so no turn is ambiguous
- Specific time estimates: rather than "this will take a few moments", estimates are concrete (e.g., "2 min")
- Make wins visible: progress is acknowledged inline so the user feels momentum
- Matter-of-fact errors: errors are stated plainly with the fix, not padded with reassurance
- Cap lists at 5 items: long lists are split or truncated to stay scannable
- No preamble, recap, or closers: removes "let me think about this", recaps, and "I hope this helps"
Installation
Claude Code:
claude plugin marketplace add ayghri/i-have-adhd
claude plugin install i-have-adhd@i-have-adhd
Codex:
codex plugin marketplace add ayghri/i-have-adhd
codex plugin add i-have-adhd@i-have-adhd
For other coding agents, see the INSTALL.md file in the repository for platform-specific instructions.
Quick Start
Once installed, invoke the skill with the /i-have-adhd slash command (Claude) or $i-have-adhd (Codex), or let the agent apply the formatting implicitly when the task benefits from the style.
Before:
Let me think about this... Your auth flow has a few moving pieces. The JWT handling seems reasonable but I noticed a potential issue with the token refresh logic. Here are my observations:
After:
Run
npm install jsonwebtoken@latest, then editsrc/auth.ts:42.
Core Concepts
- Output discipline: the skill is not a reasoning improvement; it is a formatting enforcement layer on top of whatever the agent would say
- Forkable rules: users can fork the repository and edit the
SKILL.mdfile to adjust formatting rules to their personal style - Multi-agent support: each supported agent (Claude Code, Codex, etc.) has its own installation path, but the formatting rules are shared
Advanced Features
- Custom formatting rules by forking the repo and modifying
SKILL.md - Works implicitly — no need to invoke the command on every turn if the agent picks it up
- Complements any existing agent workflow; it does not replace tooling or capabilities
Pros
- ✅ Drastically reduces token waste on verbose output
- ✅ Makes coding agent responses immediately actionable
- ✅ Works across multiple coding agents (Claude Code, Codex, etc.)
- ✅ Large and active community (16K+ stars, 893 forks)
- ✅ Free, MIT-licensed, easily forkable
Cons
- ❌ Formatting-only; does not improve agent reasoning or correctness
- ❌ Opinions on style may not suit all users or workflows
- ❌ Relies on plugin infrastructure that varies between agents
- ❌ Early-stage project; installation paths may shift as agent platforms evolve
When to Use
Use i-Have-ADHD when you work with coding agents on a daily basis and find yourself scrolling through pages of preamble to find the actual instruction. It is especially valuable in CLI-based coding workflows (Claude Code, Codex) where screen real estate is limited and fast iteration matters more than prose.