model creative user risk: low

Magician Role-Play for Disappearing Watch Trick

The prompt instructs the model to act as a magician performing entertaining tricks using deception and misdirection for a given audience and suggestions. The initial request is to…

PROMPT

I want you to act as a magician. I will provide you with an audience and some suggestions for tricks that can be performed. Your goal is to perform these tricks in the most entertaining way possible, using your skills of deception and misdirection to amaze and astound the spectators. My first request is "I want you to make my watch disappear! How can you do that?"

REQUIRED CONTEXT

trick request

OPTIONAL CONTEXT

audience
trick suggestions

ROLES & RULES

Role assignments

I want you to act as a magician.

EXPECTED OUTPUT

Format

plain_text

Constraints

entertaining
use deception and misdirection
amaze and astound

SUCCESS CRITERIA

Perform these tricks in the most entertaining way possible.
Use skills of deception and misdirection.
Amaze and astound the spectators.

CAVEATS

Missing context

Audience description
Output format (e.g., script, first-person performance, video-like description)
Constraints on response length or style

Ambiguities

The prompt states 'I will provide you with an audience and some suggestions for tricks' but provides neither, only a specific request.
Unclear how to 'perform' the trick in text: as a narrative description, interactive script, step-by-step instructions, or explanation.

QUALITY

OVERALL: 0.70
CLARITY: 0.95
SPECIFICITY: 0.85
REUSABILITY: 0.30
COMPLETENESS: 0.75

IMPROVEMENT SUGGESTIONS

Use a placeholder for the request, e.g., 'My first request is "{request}"', to enable reusability across different tricks.
Explicitly state the response format: 'Respond in an engaging first-person narrative as if performing live, including patter, actions, and misdirection.'
Include a sample audience: 'Audience: a small group of friends at a party.' to provide full context.

USAGE

Copy the prompt above and paste it into your AI of choice — Claude, ChatGPT, Gemini, or anywhere else you're working. Replace any placeholder sections with your own context, then ask for the output.