How i‑have‑ADHD Mode Numbers Multi‑Step Tasks: A Strict Formatting Guide

i‑have‑ADHD mode enforces a four-rule system for numbering multi-step tasks: use numbered lists, avoid chained actions, minimize step count, and prioritize immediate executability.

The i-have-adhd constraint system, developed by ayghri/i-have-adhd, applies specialized formatting rules designed to reduce cognitive load for users with ADHD. When tasks require multiple actions, the approach is codified in SKILL.md under the dedicated "Number multi-step tasks" rule. This specification ensures that complex workflows remain tractable without overwhelming working memory.

The Four Core Rules for Numbering Multi‑Step Tasks

Located at skills/i-have-adhd/SKILL.md (lines 42–56), the rule establishes mandatory formatting for any task sequence:

1. Write a Numbered List with Single Actions

Each list item must represent one bounded, completable action. The format uses standard markdown numbering with periods, not dashes or nested structures.

2. Eliminate Chained Actions

Steps containing "and then" clauses are prohibited. A step like "Open the file and then find the function" fails validation because it embeds multiple operations into a single item.

3. Use the Minimum Viable Step Count

Remove redundant or trivial steps. Actions that naturally belong together—such as opening a file to edit a specific line—should be combined rather than separated artificially.

4. Prioritize Immediate Executability

Every step must be actionable without additional planning. Preferred phrasing includes concrete verbs: "Open src/auth.ts", "Replace lines 42–58", or "Run npm test".

Correct vs. Incorrect Examples

The SKILL.md specification includes concrete examples demonstrating compliance and violation.

Valid three-step sequence:

1. Open `src/auth.ts`

2. Replace the `verifyToken` function (lines 42–58) with the snippet below

3. Run the test suite: `npm test -- auth.spec.ts`

Invalid sequence (violates multiple rules):

1. Open the file, find the function, replace it, and then run the tests.

The invalid example fails because it: chains four distinct actions with commas and "and then", lacks concrete file paths or line numbers, and provides no runnable command isolating the relevant test.

Why This Formatting Approach Matters

The numbering system addresses working memory limitations common in ADHD. According to the source code analysis of SKILL.md, the design intent is to let users:

• See progress at a glance through sequential numbering
• Execute without holding an extended plan in mind
• Resume interrupted workflows by identifying the last completed step

Each numbered item functions as a checkpoint, reducing the cognitive overhead of task switching and uncertainty about next actions.

Implementation Across Repository Files

The i-have-adhd skill definition appears in two synchronized locations:

File Path	Purpose
skills/i-have-adhd/SKILL.md	Canonical rule definition (lines 42–56)
.cursor/skills/i-have-adhd/SKILL.md	Cursor IDE compatibility copy

Both files maintain identical content, ensuring consistent behavior whether the mode is invoked through direct skill loading or Cursor's skill system.

Summary

• Numbered lists only — dashes or paragraphs for multi-step tasks violate the specification
• One action per step — "and then" constructions are explicitly prohibited
• Minimal steps — fold trivial actions rather than expanding the list
• Executable phrasing — every step must be immediately performable

These constraints, encoded in SKILL.md, transform potentially overwhelming task descriptions into manageable, sequential workflows.

Frequently Asked Questions

What file defines the numbering rules for multi-step tasks in i-have-ADHD mode?

The rules are defined in skills/i-have-adhd/SKILL.md at lines 42–56, under the heading "### 2. Number multi-step tasks". A Cursor-compatible copy exists at .cursor/skills/i-have-adhd/SKILL.md.

Can I use bullet points instead of numbered lists for multi-step tasks?

No. The specification mandates numbered lists exclusively for multi-step tasks. Bullets are reserved for single-step items or collections of independent options, not sequential procedures.

Why does the rule prohibit "and then" in step descriptions?

Chained descriptions like "do X and then Y" split attention across multiple incomplete actions, forcing the reader to maintain sub-goals in working memory. The rule enforces atomic steps so each item represents a single, completable unit of work.

How do I know if a step is "immediately executable"?

A step passes the executability test if it contains specific, bounded actions with concrete references: file paths (src/auth.ts), line numbers (42–58), or exact commands (npm test -- auth.spec.ts). Vague directives like "fix the issue" or "update accordingly" fail this criterion.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}