Troiana Prompt Library · Claude · Design

Reason about a destructive action's safeguards

Work out how much friction a dangerous action deserves and where it should sit.

What this prompt helps you do

A safeguard design with the confirmation approach, undo strategy, and the reasoning about friction versus safety.

When to use it

An interface allows an action that is expensive or impossible to reverse.

What to prepare

The goal, relevant source material, constraints, examples, and the decision or action this should support.

How to get a strong result

Prefer undo over confirmation where the system allows it; confirmation dialogs are dismissed reflexively.

How to use this prompt

Replace each placeholder with accurate context. Paste the complete prompt into Claude, review the response critically, resolve important questions, and use it as a considered input to the next decision—not as an unchecked final answer.

Browse all ChatGPT, Claude, and Codex prompts