Jump to content

ChatGPT: Difference between revisions

From Saxton Publishing Technical Documentation
Created page with "= High-Leverage Prompts Identified by GPT-5.2 = This page documents a set of prompts identified by GPT-5.2 as the most effective across prior conversations. These prompts consistently produced structured, reusable output, especially after deliberate prompt restructuring. Time span covered: approximately the last 3–4 months of conversations, beginning with early work on documentation structure, research notes, and site architecture. These prompts work not because the..."
 
Blanked the page
Tag: Blanking
 
Line 1: Line 1:
= High-Leverage Prompts Identified by GPT-5.2 =


This page documents a set of prompts identified by GPT-5.2 as the most effective across prior conversations. These prompts consistently produced structured, reusable output, especially after deliberate prompt restructuring.
Time span covered: approximately the last 3–4 months of conversations, beginning with early work on documentation structure, research notes, and site architecture.
These prompts work not because they are clever, but because they impose constraints, define audience expectations, and reduce ambiguity.
== 1. Knowledge Base Page Creation ==
'''Prompt'''
Create a page intended for a technical knowledge base.
Write in clear, plain American English.
Assume the reader is technical but busy.
Do not explain what a wiki is.
Use short paragraphs, clear section headers, and natural internal links.
This page should act as a foundation that other pages link from and link back to.
Avoid filler, marketing language, and meta commentary.
'''Why this works'''
* Explicitly defines the audience’s level of knowledge.
* Establishes the structural role of the page.
* Prevents common failure modes such as introductory fluff and marketing tone.
== 2. Targeted Fixes Without Regeneration ==
'''Prompt'''
Do not regenerate.
Do not explain what you did wrong.
Keep all existing structure unless I explicitly tell you to change it.
Fix only the issues I listed.
Output something I can copy.
'''Why this works'''
* Forces minimal, controlled edits.
* Prevents scope creep.
* Treats structure as authoritative rather than optional.
== 3. Case Study Restructuring With Fixed Structure ==
'''Prompt'''
Use exactly the structure in the example slide.
One slide only.
No invented metrics.
Convert this into actionable bullet points.
Maximum five bullets.
Short, concrete language.
Title must describe what is being done, not sound clever.
'''Why this works'''
* Removes ambiguity around format.
* Blocks the model from inventing data.
* Prioritizes clarity over style.
== 4. SEO Placement Without Forced Insertion ==
'''Prompt'''
Do not insert these words directly.
Instead, tell me which existing sentence could naturally include each term.
If none fit, say so.
Suggest a rewrite only if necessary.
'''Why this works'''
* Separates analysis from execution.
* Preserves editorial integrity.
* Avoids awkward keyword stuffing.
== 5. Technical Explanation With Audience Control ==
'''Prompt'''
Explain this as if the reader is technical but not academic.
No buzzwords.
No triads.
Short sentences.
Assume the reader already knows the basics and wants clarity, not onboarding.
'''Why this works'''
* Calibrates tone and depth.
* Eliminates performative explanations.
* Keeps focus on clarity rather than pedagogy.
== 6. Architecture and Strategy Validation ==
'''Prompt'''
Don’t give me steps yet.
Tell me whether this structure makes sense.
Point out what will break later.
Assume this has to scale.
'''Why this works'''
* Shifts the task from instruction to evaluation.
* Surfaces long-term risks early.
* Encourages systems-level reasoning.
== 7. Hard Constraint Output Control ==
'''Prompt'''
Output only what I can copy.
No commentary.
No explanations.
No apologies.
Match the format exactly.
'''Why this works'''
* Eliminates non-essential text.
* Forces strict compliance with format.
* Produces immediately usable output.

Latest revision as of 00:13, 15 December 2025