Question 1

What are the prompt engineering techniques?

Accepted Answer

Prompt engineering techniques are named patterns for structuring instructions to a language model. They include zero-shot, few-shot, chain of thought, retrieval-augmented generation, ReAct, self-consistency, tree of thoughts, and dozens more. This reference catalogs sixty of them, grouped into ten families: framing, structure, demonstration, reasoning, decomposition, retrieval, agent, verification, constraint, and meta.

Question 2

How many prompt engineering techniques are there?

Accepted Answer

Prompt engineering techniques number in the dozens. The Prompt Report (Schulhoff et al., 2024) catalogs 58 text-based techniques across six categories; this reference organizes 60 named techniques as a visual catalog across ten families of six.

Question 3

What is the best prompt engineering technique?

Accepted Answer

No single technique wins everywhere. Chain of Thought (Wei et al., 2022) remains the most-cited. In production, practitioners combine 3 to 5 techniques per prompt. The ten families on this reference help you pick the right combination for the task: framing, structure, demonstration, reasoning, decomposition, retrieval, agent, verification, constraint, and meta.

Question 4

What is chain of thought prompting?

Accepted Answer

Chain of Thought (CoT) prompting asks the model to reason step by step before giving a final answer. Introduced by Wei et al. in 2022, it produces large accuracy gains on math, logic, and multi-step tasks. Combine with Self-Consistency, Few-shot, and Self-Verify for production-grade reasoning prompts.

Question 5

What is few-shot prompting?

Accepted Answer

Few-shot prompting gives the model two to five input-output examples before the real task, so it can pattern-match. It is the most reliable quality-per-token gain available for most tasks. Pairs well with Chain of Thought, One-shot, and Contrastive Pairs.

The Periodic Table of Prompt Techniques.