Prompt Engineering Techniques — 20+ Methods

What it is

This is a comprehensive collection of prompt engineering techniques covering 20+ methods with runnable implementations. The repository includes techniques ranging from foundational approaches like Chain-of-Thought (CoT) and few-shot prompting to advanced strategies like ReAct, Tree-of-Thought, and self-consistency. Each technique comes with a notebook you can run directly.

The collection is designed for developers, researchers, and AI practitioners who want to move beyond basic prompting. Whether you are building AI agents, optimizing LLM outputs, or studying prompt design patterns, this resource provides concrete implementations rather than just theory.

How it saves time or tokens

Structured prompting techniques reduce wasted tokens on malformed outputs. Chain-of-Thought prompting, for example, improves accuracy on reasoning tasks, meaning fewer retry cycles. ReAct patterns combine reasoning and action in a single prompt structure, eliminating the overhead of separate planning and execution steps. By selecting the right technique for your task, you avoid the trial-and-error of discovering effective patterns from scratch.

How to use

1. Browse the technique catalog and select the method that matches your task type (reasoning, classification, generation, tool use).

1. Copy the template and adapt it to your specific domain:

# Chain-of-Thought (CoT) template
Solve the following problem step by step.

Problem: [your problem here]

Let's think through this step by step:
Step 1: ...
Step 2: ...
Step 3: ...

Therefore, the answer is: ...

1. Run the associated notebook to see the technique in action with real LLM calls and compare output quality.

Example

ReAct pattern combining reasoning and action:

Question: What is the population of the capital of France?

Thought: I need to find the capital of France first, then look up its population.
Action: Search('capital of France')
Observation: Paris is the capital of France.
Thought: Now I need to find the population of Paris.
Action: Search('population of Paris 2024')
Observation: The population of Paris is approximately 2.1 million.
Thought: I have the answer.
Answer: The population of Paris, the capital of France, is approximately 2.1 million.

This pattern grounds the model's reasoning in observable actions, reducing hallucination on factual queries.

Related on TokRepo

• Prompt Library — Browse curated prompts organized by task type and model
• AI Tools for Coding — Coding assistants that benefit from advanced prompting techniques

Common pitfalls

• Chain-of-Thought increases token usage. Use it for complex reasoning tasks where accuracy matters, not for simple classification where a direct answer suffices.
• Tree-of-Thought requires multiple LLM calls per decision point. Budget for 3-5x token cost compared to single-pass prompting.
• Not all techniques work equally well across models. Test your chosen technique on your specific model before committing to it in production.