Browse queriesAskGL
Prompt from Example
Reverse-engineer a prompt from example inputs and outputs
- Creates
- Json
- Setup
- Instant
- Starting points
- 1 version
Ready-to-use query
Change the subject, details, or style before you run it.
/write
You are an expert at reverse-engineering prompts from examples.
Given example inputs and their corresponding outputs, deduce the prompt that would produce these results consistently.
Example Input:
{{exampleInput}}
Example Output:
{{exampleOutput}}
Additional Context: {{context}}
Analyze the transformation from input to output and create a prompt that would:
1. Consistently reproduce similar transformations
2. Handle variations of the input
3. Maintain the same style/format as the example output
Return your response in this JSON format:
{
"analysis": {
"transformation": "What the prompt does to transform input to output",
"patterns": ["Patterns identified in the output"],
"style": "Detected writing style/format",
"keyElements": ["Key elements that must be preserved"]
},
"prompt": {
"title": "Descriptive title",
"description": "What this prompt does",
"systemPrompt": "The complete deduced prompt with {{variables}}",
"variables": [
{"name": "var", "type": "text", "description": "desc", "required": true}
],
"outputType": "detected output type"
},
"confidence": 0.85,
"testSuggestions": ["Suggested test inputs to verify the prompt works"]
}.
Be specific about intent, audience, and desired output quality.
Keep structure clear and directly useful without filler.
!Be concrete and specific; avoid vague language.
!Return only the final output.
>json