one-shot prompting

**One-shot prompting** is the **prompting strategy that provides exactly one demonstration example before the target task** - it offers a lightweight way to steer output format and behavior when context is limited. **What Is One-shot prompting?** - **Definition**: Single-example in-context prompt that illustrates desired mapping or response structure. - **Use Objective**: Give enough guidance to reduce ambiguity while minimizing token overhead. - **Common Scenario**: Structured outputs such as JSON, classification labels, or templated summaries. - **Performance Profile**: Usually stronger than zero-shot for format adherence, but less robust than few-shot on complex tasks. **Why One-shot prompting Matters** - **Token Efficiency**: Delivers meaningful steering with minimal prompt length increase. - **Format Reliability**: A single concrete example often improves schema compliance significantly. - **Fast Iteration**: Easy to update and test during application development. - **Cost Control**: Lower context use helps manage latency and inference cost at scale. - **Operational Simplicity**: Useful default when full few-shot context is unavailable. **How It Is Used in Practice** - **Example Selection**: Choose a representative example with clear structure and no ambiguity. - **Instruction Pairing**: Combine concise rules with the one-shot demonstration. - **Validation Checks**: Test against edge cases to confirm the single example generalizes adequately. One-shot prompting is **an efficient middle ground between zero-shot and few-shot prompting** - it provides targeted guidance with low token cost and strong practical utility in production systems.