PromptsApr 1, 2026·2 min read

Prompt Engineering Techniques — 20+ Methods

Comprehensive prompt engineering collection with 20+ techniques. 7.3K+ stars. Chain-of-Thought, ReAct, Tree-of-Thought, runnable notebooks.

PR
Prompt Lab · Community
Quick Use

Use it first, then decide how deep to go

This block should tell both the user and the agent what to copy, install, and apply first.

Pick a technique, copy the template, adapt to your task:

Chain-of-Thought (CoT):

Solve this step by step:
1. First, identify the key variables
2. Then, set up the equation
3. Finally, solve and verify

Problem: [your problem here]

ReAct (Reasoning + Acting):

Answer the following question by alternating between Thought, Action, and Observation steps.

Question: [your question]

Thought 1: I need to...
Action 1: Search for...
Observation 1: [result]
Thought 2: Based on this, I should...

Self-Consistency:

Generate 5 different solutions to this problem, then select the most common answer:

Problem: [your problem]

Solution 1: ...
Solution 2: ...
[Let the model generate multiple paths, then pick the majority]
Intro

This is a curated collection of 20+ prompt engineering techniques, each with explanations, templates, and runnable Jupyter notebooks. Covers:

Foundational:

  • Zero-Shot Prompting — No examples needed, just clear instructions
  • Few-Shot Prompting — Provide examples for the model to follow
  • Chain-of-Thought (CoT) — Step-by-step reasoning
  • Role Prompting — Assign an expert persona

Intermediate:

  • ReAct — Combine reasoning with tool use
  • Self-Consistency — Multiple reasoning paths, majority vote
  • Tree-of-Thought — Explore multiple solution branches
  • Least-to-Most — Break complex problems into sub-problems

Advanced:

  • Constitutional AI Prompting — Self-critique and revision
  • Meta-Prompting — Prompts that generate prompts
  • Automatic Prompt Engineering (APE) — Let the model optimize its own prompts
  • Prompt Chaining — Sequential prompts where output feeds input
  • Skeleton-of-Thought — Generate outline first, then fill in details

Each technique includes: concept explanation, when to use it, template, Python notebook, and real-world examples.

FAQ

Q: Which technique should I start with? A: Chain-of-Thought for reasoning tasks, Few-Shot for format-sensitive tasks, ReAct for tasks requiring external tool use.

Q: Do these work with all LLMs? A: Yes. These are model-agnostic techniques. They work with GPT-4, Claude, Gemini, Llama, Mistral, and any instruction-following LLM.

Q: Can I use these in production? A: Absolutely. Each technique is production-tested. The notebooks include evaluation metrics so you can measure effectiveness.

Works With

  • Any LLM: Claude, GPT-4, Gemini, Llama, Mistral
  • Claude Code, Cursor, Codex (for coding-specific prompts)
  • LangChain, LlamaIndex (for chained prompts)
  • Jupyter Notebook (runnable examples)
🙏

Source & Thanks

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets

Cursor vs Claude Code vs Codex — AI Coding Compared

In-depth comparison of the three leading AI coding tools in 2026. Covers pricing, context window, MCP support, agent capabilities, and best use cases for each platform.

Prompt Lab

Structured Outputs — Force LLMs to Return Valid JSON

Complete guide to getting reliable structured JSON from LLMs. Covers OpenAI structured outputs, Claude tool use, Instructor library, and Outlines for guaranteed valid responses.

Prompt Lab

LitGPT — Fine-Tune and Deploy AI Models Simply

Lightning AI's framework for fine-tuning and serving 20+ LLM families. LitGPT supports LoRA, QLoRA, full fine-tuning with one-command training on consumer hardware.

Prompt Lab
◈Home🔍Search👤Me