Few Shot Example Designer
Craft high-signal, pedagogically effective examples that teach an AI model exactly how to perform a specific task.
- Category
- AI & Prompt Engineering
- Output
- Markdown
- Fields
- 3
- prompt-engineering
- content-structuring
- ai-communication
- training
- output-optimization
Prompt
You are an expert prompt engineer specializing in in-context learning and few-shot demonstration design. Your goal is to craft high-signal, pedagogically effective examples that teach an AI model exactly how to perform a specific task.
TASK:
Design a set of few-shot examples for the following task: Task description.
CONTEXT & CONSTRAINTS:
- Target model: Target model (If left blank, apply general best practices for modern LLMs).
- Desired output format: Output format (If left blank, use plain text).
- You must teach the boundaries and edge cases, not just the happy path.
- Use only the information provided. Do not invent facts, names, scenarios, or technical details. If any critical detail is missing or ambiguous, list it under "Open Questions" before generating examples.
- Ensure strict consistency in formatting across all examples, as formatting consistency is a core learning signal.
INSTRUCTIONS:
1. Analyze the task to identify input variety: easy, typical, and hard/edge cases.
2. Create exactly 3 to 5 input-output pairs. Each pair must clearly show the exact input and the precise, correctly formatted output.
3. Include at least one example that demonstrates how to handle ambiguity, conflicting instructions, or tricky constraints.
4. Add a concise instruction preamble (1-2 sentences) that sets expectations and pairs naturally with the examples.
5. Provide brief notes on example ordering and the pedagogical reasoning behind your choices.
FORMAT-SPECIFIC RULES:
- If Output format is "JSON": Return only valid JSON with no markdown fences or surrounding text. Use keys like "input", "output", and "notes".
- If Output format is "Markdown table": Return a single markdown table with columns for Input, Output, and Edge Case Notes.
- If Output format is "Code block": Wrap each example in a language-appropriate code block with clear input/output markers.
- If Output format is "Step-by-step list": Use numbered lists for inputs and bulleted lists for outputs, with clear separators.
- For any other format or if left blank, use clear "Input:" and "Output:" labels with consistent spacing.
OUTPUT STRUCTURE:
Return your response in the following order:
1. Instruction Preamble
2. Few-Shot Examples (formatted according to Output format)
3. Ordering & Design Notes
4. Open Questions (only if information is missing or ambiguous)
How to use this prompt
- Fill in the fields above. Required ones are marked with *.
- Check the prompt preview: your answers appear highlighted in place.
- Select Copy prompt and paste it into ChatGPT, Claude, Gemini or any AI assistant.
More AI & Prompt Engineering prompts
- Structure and Format AI OutputTransform unstructured input into a polished, professional output that strictly matches the requested target format.
- Automate AI WorkflowDesign a reliable, step-by-step AI automation workflow based on the user's specifications.
- Image Generation PromptTransform a user's core concept into highly optimized, ready-to-use prompts for AI image generation.
- LLM Evaluation and Model ComparisonYou are an expert AI model evaluator and technical consultant.
- Prompt Engineering Deep OptimizationRewrite the provided draft prompt using advanced prompt engineering best practices, making it more precise, reliable, and structured.
- System Prompt Production ReadyYou are an expert prompt engineer tasked with crafting production-ready system instructions for AI assistants.