This project reformats coding agent responses to be concise and action-first by leading with the next step, numbering instructions, and removing preamble. It is designed for users who find verbose agent outputs difficult to parse and integrates via plugin marketplaces.
Project overview
It applies ten specific formatting rules to agent outputs, adapting psychological principles from The Adult ADHD Tool Kit specifically for LLM response formatting.
Project type
AI Coding · Prompt Engineering
Use cases
Coding & Development
Deployment
Refer to project documentation
License
MIT
Best for
Developers or general users who want their coding agent outputs formatted with action-first, numbered steps and suppressed tangents.
Key capabilities
Modifies agent output to lead with the next action and place steps first, removing preamble, tangents, and closers such as "Hope this helps!".
Applies 10 rules including numbered steps, one next step, suppressed tangents, state restatement, specific time estimates, and capped lists.
Allows users to fork the repo, edit the skill rules file, and install their own customized version.
Limitations and risks
The effectiveness of the formatting depends entirely on the underlying coding agent correctly interpreting and applying the 10 rules.
Getting started
Installation difficulty is rated easy because it uses a single marketplace command and plugin install command with no local clone needed. The first success path involves adding the marketplace, installing the plugin, and typing the slash command to activate it.
Evidence and sources
GitHub project description: A skill for your coding agent to stop it from burying the answer. ADHD-friendly output.
README: ```bash claude plugin marketplace add ayghri/i-have-adhd claude plugin install i-have-adhd@i-have-adhd ``` Then type `/i-have-adhd`. No local clone needed — Claude Code fetches th…
README: A skill for your coding assistant that stops it from burying the answer. Action first. Steps numbered. No "Hope this helps!"
README: > 1. Open `src/auth.ts` > 2. Replace `verifyToken` (lines 42–58) with the snippet below > 3. Run `npm test -- auth.spec.ts`
README: 1. Lead with the next action. 2. Number multi-step tasks. 3. End with one concrete next step. 4. Suppress tangents. 5. Restate state every turn. 6. Specific time estimates (minute…