How a Simple Skill.md Earned 18K Stars by Making AI Say the Answer First

Repository

GitHub repository https://github.com/ayghri/i-have-adhd provides a Skill.md that defines how AI coding assistants should structure their responses.

Core rules (10 rules, three essential ideas)

State the answer first – the AI should immediately give the command to run or file to edit.

List steps in order – enumerate each step sequentially and indicate progress.

Show only useful content – remove irrelevant background, polite filler, and repeated summaries.

The skill also requires the AI to report current progress on longer tasks and to provide explicit error locations, possible causes, and remediation steps when failures occur.

Before‑after comparison

Without the skill, an AI response to an authentication error first praises the issue, discusses middleware, token verification, and cookies, then mentions a possible change in src/auth.ts and suggests checking other dependencies. With the skill, the response jumps directly to the actionable fix, presents a concise numbered step list, and ends with a clear next action.

Installation and usage

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 --ref main
codex plugin add i-have-adhd@i-have-adhd

After installation, invoke the skill in a session with /i-have-adhd. To load the rules automatically at every session start, create a marker file: touch ~/.claude/.i-have-adhd-always or add the rules to ~/.codex/AGENTS.md for Codex.

Safety rules

Dangerous operations such as file deletion require explicit confirmation. After three consecutive failures, the AI stops retrying and re‑examines its previous judgments.

Applicability

The skill is suitable for bug fixing, project deployment, and multi‑step tasks, while allowing temporary disabling when detailed technical explanations are needed.