Twenty-Two Prompting Techniques, Ready to Run

What it does This repository holds 22 hands-on Jupyter Notebook tutorials that demonstrate prompt engineering techniques from basic templates to advanced strategies. Each notebook isolates a specific method—few-shot learning, chain-of-thought, self-consistency, prompt chaining, and others—with runnable code rather than abstract theory. It functions as an executable reference for developers who want to observe how different prompting patterns directly affect LLM output.

The interesting bit The project treats prompt engineering as a learnable discipline rather than dark art, organizing techniques into five progressive categories: Fundamental Concepts, Core Techniques, Advanced Strategies, Advanced Implementations, and Optimization. The author is building a broader GenAI curriculum around it, with companion write-ups, a commercial book, and sibling repositories covering RAG and agents.

Key highlights

• 22 runnable notebooks covering techniques from basic structures to tree-of-thought and constrained generation
• Clear taxonomy with five categories and direct links to each .ipynb file
• Free companion write-ups on DiamantAI that add intuition and examples beyond the notebooks
• Community-driven, accepting contributions via Discord and CONTRIBUTING.md
• Explicitly connected to related repos on RAG, GenAI agents, and agent memory techniques

Caveats

• The README contains prominent promotions for the author’s book, newsletter, and hiring collective; the technique index requires scrolling past several marketing sections
• It is unclear from the README whether running the notebooks requires paid API keys or specific local model infrastructure

Verdict Useful for developers and data scientists who need a structured, code-first on-ramp to systematic prompting. Less interesting if you already have a production prompt library and are looking for framework-specific implementation details.