i-have-adhd Makes Your Coding Agent Lead With the Answer

Your coding assistant answers a one-line question with three paragraphs of tour before the actual command shows up. i-have-adhd — a 48k-star skill — flips that around: the next action comes first, multi-step work is numbered, and every reply ends with one concrete step. No ADHD diagnosis is needed, the README says so up front; it is an output style anyone can switch on.

Why This Skill Matters

Ten rules drive the behavior (full text in skills/i-have-adhd/SKILL.md): lead with the next action, number multi-step tasks, end with one concrete next step, suppress tangents, restate state every turn, give time estimates in concrete units rather than "a bit", make wins visible, keep errors matter-of-fact, cap lists to 5 items, and drop preambles, recaps, and closers.

The README's before-and-after example makes the change concrete. Faced with an outdated jsonwebtoken call, the styled reply is:

Run npm install jsonwebtoken@latest, then edit src/auth.ts:42.

1. Open src/auth.ts
2. Replace verifyToken (lines 42–58) with the snippet below
3. Run npm test -- auth.spec.ts

Next: paste the first failing line if any test fails.

The repo also documents per-platform setup for Claude Code, Codex, Gemini CLI, GitHub Copilot, OpenCode, Pi, Zed, and other harnesses, plus an optional always-on mode for most of them.

Installation

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

Type /i-have-adhd in a session. If the command is not in autocomplete yet, restart the agent — the plugin index is read at startup.

Any harness that reads Agent Skills can take the CLI route instead:

npx skills add ayghri/i-have-adhd

Real Workflow: Invoke, Then Test the Style

1. Start a new session and type /i-have-adhd.
2. Ask a real question from your project.
3. Check the reply: command or file path first, numbered steps, no warm-up.

Codex users invoke with $i-have-adhd instead — the repo notes Codex will not activate the skill automatically. In Claude Code, Qwen Code, and Grok the skill sets disable-model-invocation: true, so nothing changes until you invoke it. Rules stay on for the session; say "stop adhd mode" or "normal mode" to switch back.

Real Workflow: Always-On Mode

If you want the style on every session without invoking anything:

touch ~/.claude/.i-have-adhd-always

A SessionStart hook then loads the full ruleset from message one, every session. The flag file is the switch: delete it to return to on-demand. The always-on ruleset also carries exceptions — explain fully when asked to explain, confirm before destructive actions, and after three failed fixes stop and name the doubtful assumption.

Tips

• Installed but the style still drifts? Open a new session; skills are indexed at session start.
• Tune the rules by forking: edit skills/i-have-adhd/SKILL.md, then swap your copy in via the marketplace commands — drop the upstream copy first, because fork and upstream share both names.
• claude plugin marketplace add wants the owner/repo form; a local path must point at the repo root.
• The rules are loosely based on The Adult ADHD Tool Kit by J. Russell Ramsay and Anthony L. Rostain, adapted for how an LLM should respond.

When Not to Use This

This is a style skill, not a capability upgrade — it will not fix bugs or write better code; it only changes how replies are shaped. When you actually want the long tour (learning a codebase, writing onboarding docs), invoke it selectively or skip it: the exceptions cover being asked to explain, but a permanent always-on mode works against exactly those sessions.

See the leaderboard for more skills.