WILD CORPUS · reddit
PQS 56 (C) - prompt from www.reddit.com
Source: www.reddit.com · Scraped 2026-05-04 · Scored 2026-05-04
Score
C56 / 80
gemma4:latest · local · pqs-v2.0 · canonical
Clarity9 / 10
Specificity8 / 10
Context7 / 10
Constraints6 / 10
Output format7 / 10
Role definition8 / 10
Examples3 / 10
CoT structure8 / 10
The prompt
Everyone is trying to sell you a $297 “Prompt Engineering Masterclass” right now. but 90% of that stuff is recycled fluff wrapped in a Canva slideshow. Let me save you time (and your wallet): The best prompt isn’t even a prompt. It’s a meta-prompt. It doesn’t just ask AI for an answer—it tells AI *how to be better at prompting itself*. Here’s the killer template I use constantly: # The Pro-Level Meta-Prompt Template: Act as an expert prompt engineer. Your task is to take my simple prompt/goal and transform it into a detailed, optimized prompt that will yield a superior result. First, analyze my request below and identify any ambiguities or missing info. Then, construct a new, comprehensive prompt that. 1. Assigns a clear Role/Persona (e.g., “Act as a lead UX designer...”) 2. Adds Essential Context so AI isn’t just guessing 3. Specifies Output Format (list, table, tweet, whatever) 4. Gives Concrete Examples so it knows your vibe 5. Lays down Constraints (e.g., “Avoid technical jargon,” “Keep it under 200 words,” etc.) Here’s my original prompt: **\[Insert your basic prompt here\]** Now, give me only the new, optimized version. # You’re giving the AI a job, not just begging for an answer. * It forces clarity—because AI can’t improve a vague mess. * You get a structured, reusable mega-prompt in return. * Bonus: You start learning better prompting by osmosis. Prompt engineering isn’t hard. It’s just about being clear, clever and knowing the right tricks
This prompt was scraped from a public source. The score reflects the input as written, not the quality of any output it produced. The AI input quality problem is the gap between what people type and what the model can act on.