Prompt Engineering Playbook 📜✨

Live site: https://your-org.github.io/prompt-engineering Boost your LLM projects with ready‑to‑use prompts, proven design patterns, and evaluation tools (PT‑BR 🇧🇷 & EN 🇬🇧).

---

📌 Why this repo?

Crafting great prompts is half art, half science – and the best examples are scattered across the internet. This repository collects battle‑tested prompts, explanatory patterns, and automation scripts so you can:

1. Copy & paste high‑performing prompts for generation, extraction, classification, and transformation tasks.
2. Understand the underlying patterns (Chain‑of‑Thought, ReAct, Persona Switching, etc.).
3. Evaluate different models & settings using an open‑source harness wired into CI.
4. Browse everything online via a GitHub Pages site with search, sidebars, and versioning.

---

🌐 Live Docs on GitHub Pages

The Markdown in this repo is rendered as a static site built with Docusaurus 3.

Action	Command
Run docs locally	bun run docs:dev
Build static site	bun run docs:build
Preview build	bun run docs:serve

A GitHub Action (.github/workflows/docs.yml) automatically deploys the /build output to the gh-pages branch after every push to main.

Tip: All Markdown under prompts/, patterns/, and templates/ is automatically picked up by Docusaurus via docs/plugin-content-pages.

---

🗂 Repository layout

prompt-engineering/
├─ README.md                ← you are here
├─ CONTRIBUTING.md          ← how to contribute prompts & docs
├─ LICENSE                  ← MIT by default – change if needed
├─ docs/                    ← Docusaurus config & static assets
│  ├─ docusaurus.config.ts
│  └─ sidebar.ts
├─ prompts/                 ← ready‑to‑use prompts (with YAML front‑matter)
│  ├─ generation/
│  ├─ classification/
│  ├─ extraction/
│  └─ transformation/
├─ patterns/                ← conceptual guides & anti‑patterns
├─ templates/               ← blank skeletons to kick‑start new prompts
├─ evals/                   ← evaluation harnesses & metrics
└─ tools/                   ← CLI utilities (index generator, linter, etc.)

Front‑matter convention

Every prompt starts with required YAML metadata so scripts can build an index and Docusaurus can create filters:

---
title: "LinkedIn Post Generator"
category: generation
tags: [marketing, social-media, portuguese, english]
model: agnostic
temperature: 0.8
author: "Ciro Maciel"
updated: 2025-05-17
description: |
  Generates an engaging LinkedIn post on any topic.
---

The prompt body goes below the front‑matter.

---

🚀 Quick start

# 1. Clone the repository
$ git clone https://github.com/your-org/prompt-engineering.git
$ cd prompt-engineering

# 2. Install dependencies (Node ≥18 + bun recommended)
$ bun install        # or npm install / pnpm i

# 3. Serve docs locally (auto‑reload)
$ bun run docs:dev   # then open http://localhost:3000

# 4. Explore prompts in /prompts & /patterns ✨

---

🧪 Evaluating prompts

Inside evals/ you’ll find harnesses based on Promptfoo and LangChain benchmarks. Run bun run tools/eval --prompt prompts/generation/linkedin_post.md to compare models and hyper‑parameters.

---

🤝 Contributing

We welcome pull requests! Please skim CONTRIBUTING.md, but in short:

1. Fork → Branch → Add your Markdown under the correct folder.
2. Keep front‑matter complete; run bun run check to lint.
3. Commit + open PR.
4. CI will run lint, docs build, and eval smoke‑tests.

Feel free to open issues with questions or suggestions.

---

📄 License

Released under the MIT License – see LICENSE.

---

✨ Acknowledgements

• OpenAI best‑practice research.
• Bun for ultra‑fast TypeScript tooling.
• The community of prompt engineers who share their wisdom.

Happy prompting & see you on the docs site! 🚀