Glonce

i-have-adhd skill stops coding agents from burying the answer

toolsThe Glonce editorsHacker News🔥64Glonce rating64Impact45Novelty40Relevance85Surprise25Players15Credibility80
i-have-adhd skill stops coding agents from burying the answer

A GitHub project called i-have-adhd distributes a skill and plugin for Claude Code aimed at one specific failure mode of coding agents: burying the actual answer under filler. The skill's ten rules, written out in full in its SKILL.md file, are: lead with the next action; number multi-step tasks; end with one concrete next step; suppress tangents; restate state every turn; give specific time estimates in minutes rather than vague phrases like "a bit"; make wins visible; report errors matter-of-factly; cap any list at 5 items; and drop preamble, recap and closing lines such as "Hope this helps!"

The project's own description says it is loosely based on The Adult ADHD Tool Kit, a self-help book by J. Russell Ramsay and Anthony L. Rostain, with its ideas adapted for how an LLM should respond rather than how a human should organize their day. The repository offers its instructions translated into several languages alongside English, including Chinese, Portuguese, Japanese, Vietnamese, Korean and Thai.

Installation runs through Claude Code's plugin system: pointing the CLI at the GitHub repository and following the AGENTS.md instructions installs the skill directly. The project also documents how to customize it: fork the repository, edit skills/i-have-adhd/SKILL.md, then remove the upstream plugin and marketplace entry before adding and installing the forked copy under the same plugin name, followed by a restart of Claude Code. The project is released under the MIT license.

Key facts

  • i-have-adhd is a Claude Code skill and plugin, distributed on GitHub, built around 10 explicit rules for how a coding agent should format its replies.
  • The rules include leading with the next action, numbering multi-step tasks, capping any list at 5 items, giving time estimates in minutes, and cutting preamble, recaps and closing lines like "Hope this helps!"
  • The project says it is loosely based on The Adult ADHD Tool Kit by J. Russell Ramsay and Anthony L. Rostain, adapting the book's ideas for LLM output rather than human time management.
  • Installation and customization both go through Claude Code's plugin commands, including a documented fork-and-swap procedure for editing the rules.
  • The project is released under the MIT license.

Why it matters

Coding agents built on top of chat models inherit a chat habit that gets in the way at the terminal: hedged, padded answers that open with a recap, wander through caveats, and close with a pleasantry instead of the next command. i-have-adhd names that failure mode directly and turns fixing it into a short, checkable ruleset rather than a vague style request, which is the kind of thing a plugin system can actually enforce turn after turn.

Who it affects

Anyone running Claude Code as a coding assistant and finding its replies too verbose or too indirect. Because the skill installs as a plugin rather than a model change, it affects only sessions where a user has chosen to install it, not Claude Code's default behavior.

How to use it

Install it by pointing Claude Code's plugin manager at the ayghri/i-have-adhd GitHub repository and following the setup steps in its AGENTS.md file, then invoking it as a skill. To change the rules, fork the repository, edit skills/i-have-adhd/SKILL.md, remove the upstream plugin and marketplace entry, add the fork under the same plugin name, and restart Claude Code before re-invoking it. The project carries an MIT license, so forking and modifying it is explicitly permitted.

How solid is it

The source is the project's own GitHub README plus its linked SKILL.md, describing the rules and installation steps in the author's own words; there is no independent report or benchmark backing the claims here. The listing drew substantial attention on Hacker News, but that reflects interest in the idea, not evidence that it works.

Risks and caveats

The text gives no data, testimonial or benchmark showing that the ruleset measurably changes agent output beyond the author's description; the ADHD framing is an explicit loose adaptation of a self-help book for human time management, not a clinical claim about the tool. The text also does not say whether the skill works with coding agents other than Claude Code, or name who maintains the project.