PromptCraft 🛠️

Crafted prompts for lightweight, high-impact AI agents.

This repo is a companion to the article "Simple Agents, Big Wins" — written for builders who want to create useful AI tools without drowning in complexity.

While the essay focused on mindset and simplicity, PromptCraft delivers the practical side: real prompts, real patterns, no fluff.

---

🧠 Why This Exists

You don’t need a swarm of agents or complex pipelines to build something valuable.

Most real-world agents start with:

• A single task
• A focused prompt
• A clear, useful output

But designing prompts that are consistent, clear, and reusable takes craft. PromptCraft gives you that craft — in a modular, model-agnostic format.

---

📦 What You’ll Find Here

• ✅ Prompt templates for different agent types
• 🎯 Tone + output modifiers to guide response behavior
• 🧪 Real examples you can test, remix, and apply
• 📐 Structure and clarity for anyone tired of trial-and-error

---

👤 Who It's For

• Solo devs shipping fast
• Founders prototyping AI ideas
• Designers testing agent UX
• Engineers tired of vague prompt advice

No special framework required — works with:

• Grok, ChatGPT, Claude, Gemini
• Open-source models (Mistral, LLaMA, etc.)
• LM Studio, Ollama, custom UIs
• Any LLM that responds to text input

---

⚡ How to Use It

1. Browse /templates and grab a prompt style
2. Add optional modifiers for tone or output control
3. Test directly in your model of choice
4. Remix + extend as needed

This repo is framework-free. Just copy, test, ship.

---

🧵 Read the Original Essay

"Simple Agents, Big Wins" It explains the mindset behind this repo: Start small. Stay useful. Ship what helps.

---

🔄 Always Evolving

PromptCraft will grow over time with:

• New prompt formats
• More real-world examples
• Community contributions

Got something to add or suggest? → Open an issue or PR. → Or fork, remix, and make it yours.

---

📁 Structure

PromptCraft/
│
├── README.md                 ← What this is, how to use it
│
├── /templates                ← Prompt formats for agent types
│   ├── single-shot.md
│   ├── context-injected.md
│   └── rule-bound.md
│
├── /modifiers                ← Optional prompt tone + output controls
│   ├── tone-modifiers.md
│   └── output-structure.md
│
├── /examples                 ← Ready-to-use agent prompt examples
│   ├── productivity-agent.md
│   └── story-generator.md
│
└── /tools (optional)         ← Helper scripts for testing
    └── inject_context.py