The 5‑line prompt brief
Turn vague tasks into clear, on-brand outputs with a tiny template.
You can get great results without copying sensitive data into a chat box. Use this guide to protect privacy, reduce risk, and still get useful drafts and analysis.
If you wouldn’t put it in a public Google Doc, don’t paste it into an AI tool—unless you have an explicit, approved workflow for that tool.
Even when you trust the vendor, you want to minimize exposure and keep a clean boundary between “helpful context” and “sensitive content.”
Full names (when not necessary), addresses, phone numbers, SSNs/IDs, passport numbers, student records, medical info, HR details.
Passwords, API keys, private tokens, access links, recovery codes, internal URLs with embedded credentials.
Client contracts, NDAs, pricing terms, deal notes, legal advice threads—especially anything tied to a real person or company.
Support tickets with identifying details, private roadmap notes, unreleased financials, proprietary processes, internal incident write-ups.
You can usually get the same value with sanitized inputs and tighter definitions of done.
This helps you get useful output without leaking sensitive details.
Context (sanitized): [bullets with non-sensitive facts only]
Goal: [what you want]
Constraints: No real names. No proprietary details. No legal/medical advice. Keep it concise.
Output format: [bullets / table / steps]
Definition of done: [what “good” looks like]
Better inputs and better checks.
Turn vague tasks into clear, on-brand outputs with a tiny template.
Quick checks for facts, numbers, assumptions, and sources.