# Writing Instructions

> Formatting and privacy rules for SKILL.md, agent, and command files. Load before creating, editing, or reviewing one, and pair with the skill that does the work: skill-creator:skill-creator, skill-extractor:skill-extractor, or claude-md-management:revise-claude-md / claude-md-management:claude-md-improver.

- Skill: `motlin/writing-instructions` (Agent Skill)
- Install (CLI): `npx skillmds@latest add motlin/writing-instructions`
- Raw SKILL.md: https://api.skillmd.com/api/skills/motlin/writing-instructions/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: motlin (https://skillmd.com/u/motlin)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/motlin/writing-instructions

---


# Writing Instructions for Skills, Agents, and Commands

## Use descriptive headers, not numbered steps

Never number steps (`## Step 1:`, `## Step 2:`). LLMs reorder steps during edits, and numbered headers force renumbering on every insert or delete. Name the action instead: `### Run the build`, not `### Step 1: Run the build`.

When steps must be run in sequence, say so, without numbering.

## Scrub personal data

Never put real personal data in an example: names, emails, places, phone numbers, addresses, or verbatim private text.

Rebuild the example with placeholders that keep the shape: `@Alice` and `@Bob` for people, generic or popular names and places, invented text for content.

Before committing, scan the diff for names and content you did not invent.