Prompt Engineering Technical Writing

Prompt Engineering & Technical Writing

Create effective AI prompts and technical content. Covers prompt structure, anti-patterns, cross-platform compatibility, and WordPress publishing.

Prompt Structure

Core Prompts (< 2000 tokens)

Self-contained, portable across all platforms:

# [Prompt Name]

> **Category**: [category]
> **Platforms**: All

[Prompt content — complete instructions in a single block]

## Usage
[How to use]

## Variables
- [VARIABLE]: [description]

Extended Prompts

Unlimited length, may reference skills or tools:

# [Prompt Name]

> **Category**: [category]
> **Platforms**: Claude Code | Cursor

<prompt>
[Full prompt content with tool references]
</prompt>

## Usage
[Platform-specific instructions]

## Customization
[What can be modified]

Writing Rules

Anti-Patterns (Never Use)

Banned words: delve, tapestry, myriad, leverage, utilize, seamless, robust, paradigm, synergy, cutting-edge, game-changer

Banned transitions: furthermore, moreover, additionally, subsequently, consequently

Banned openings: "In today's...", "Have you ever...", "Let's dive into..."

Good Practices

• Use active voice and present tense
• Use second person ("you")
• Use contractions (it's, don't, won't)
• Vary sentence and paragraph lengths
• Be direct — state things without hedging
• No unnecessary qualifiers (very, quite, rather)

Technical Blog Posts

When drafting from a codebase:

1. Research: Read recent commits, review changed files, note design decisions
2. Plan: Opening hook, overview, problem/value, technical details, challenges, future
3. Write: 500-1200 words, code snippets 5-15 lines, save to .blog/YYYY-MM-DD-slug.md
4. Publish (optional): Create WordPress draft via REST API

Prompt Quality Dimensions

Score each 1-5:

Cross-Platform Tips

• ChatGPT/Gemini: No tool access — prompts must be self-contained
• Claude Code: Can reference files, use skills, run commands
• Cursor: Can read project files, uses rules files
• Core prompts must work on the lowest common denominator

Token Optimization

For core prompts under 2000 tokens:

• Use concise instructions, not verbose explanations
• Prefer lists over paragraphs
• Include only essential code examples
• Reference skills instead of duplicating knowledge
• Remove redundant phrasing

References

• WordPress Documentation Style Guide
• richtabor/skills
• Anthropic Prompt Engineering Guide

Converted and distributed by TomeVault — claim your Tome and manage your conversions.