HomeLearnPrompt Engineering Glossary
Educational Guide

Prompt Engineering Glossary: Essential Terms Explained

Master prompt engineering terminology. Definitions for every term you need to know as an AI engineer.

Definition

A comprehensive reference of terms and concepts used in prompt engineering and LLM application development.

Core Concepts

**Prompt**: The input text sent to an LLM **System prompt**: Instructions that set the model's behavior **User prompt**: The actual query or request **Completion**: The model's generated response **Context window**: Maximum tokens the model can process **Temperature**: Controls randomness in outputs

Techniques

**Zero-shot**: Prompting without examples **Few-shot**: Providing examples in the prompt **Chain-of-thought**: Requesting step-by-step reasoning **Role prompting**: Assigning a persona to the model **Prompt chaining**: Connecting multiple prompts **RAG**: Retrieval-Augmented Generation

Evaluation Terms

**Ground truth**: The correct expected output **Baseline**: Reference performance to compare against **Regression**: Quality decrease after changes **Hallucination**: Made-up or incorrect information **Eval set**: Dataset for testing prompt quality

Advanced Concepts

**Fine-tuning**: Training a model on custom data **Embeddings**: Vector representations of text **Tool use**: LLMs calling external functions **Agents**: Autonomous LLM-powered systems **Guardrails**: Safety constraints on outputs

Related Topics

Prompt Regression

A prompt regression occurs when changes to a prompt or its context cause the LLM to produce lower-quality outputs than before. This can happen due to prompt edits, model updates, or changes to system context.

Prompt Testing Best Practices

A collection of proven methods and workflows for systematically testing LLM prompts to ensure quality, reliability, and safety in production applications.

LLM Evaluation Metrics

LLM evaluation metrics are quantitative and qualitative measures used to assess the quality, accuracy, and usefulness of large language model outputs.

Put This Knowledge Into Practice

Use PromptLens to implement professional prompt testing in your workflow.

Start Free