Prompt Engineering#few shot#in context learning#examples#prompt engineering#chatgpt

Few-Shot Examples Prompt

Teach the model a pattern by showing 2-5 input-output examples first.

Works with:ChatGPTClaudeGemini
Full prompt
Follow the pattern shown in the examples below.

Example 1:
Input: [input 1]
Output: [output 1]

Example 2:
Input: [input 2]
Output: [output 2]

Example 3:
Input: [input 3]
Output: [output 3]

Now, your turn:
Input: [new input]
Output:

Make sure your output matches the exact format and style of the examples.
💡
Organize your AI prompt library with Marqly

Never lose high-performing prompts in chat history. Clip prompts directly from ChatGPT, Claude, and Gemini with theMarqly extension, tag them by project, and find them withsemantic search.

How to use this prompt

Study the structure, then adapt it to your own task. The key is giving the model a clear role, format, and evaluation criteria.

  1. Copy the full prompt above.
  2. Replace the bracketed placeholders with your own details.
  3. Paste it into chatgpt, claude, gemini (or your preferred AI tool).
  4. Iterate on the output: add constraints, ask follow-ups, or combine multiple results.

Why this prompt works

It gives the AI a clear role, specific inputs, and an output format. That structure reduces vague responses and gives you a draft you can refine in minutes instead of hours.

Frequently asked questions

How many examples should I include?

2-5 diverse examples usually work best; more can waste context.

What if examples are too similar?

Diverse examples reduce bias and help the model generalize the pattern.