⚙️ PromptKit

A production-grade library for structured LLM prompt engineering.


Explore the docs »

Report Bug · Request Feature

---

Why PromptKit?

Managing prompts for Large Language Models (LLMs) can quickly become messy. Hardcoding prompts as f-strings mixes logic with presentation, lacks validation, and makes it difficult to reuse and test them.

PromptKit solves this by treating your prompts as structured, version-controlled assets. By defining prompts in simple YAML files, you get:

• Clean Separation: Your prompt templates, logic, and configuration are separate from your application code.
• Safety & Reliability: Built-in validation ensures your prompts receive the correct inputs every time.
• Reusability: Define a prompt once and use it anywhere—in your Python code or directly from the CLI.
• Clarity: A clear, human-readable format for prompts that anyone on your team can understand.

✨ Features

• 📝 Declarative & Structured: Define prompts in simple YAML files with powerful Jinja2 templating.
• 🔍 Built-in Validation: Use Pydantic schemas to validate prompt inputs before they are ever sent to the LLM.
• 🏗️ Engine Agnostic: A clean engine abstraction layer supports OpenAI, with Ollama and other local models on the way.
• 💰 Cost & Token Estimation: Estimate token counts and potential costs before executing a prompt.
• 🖥️ Powerful CLI: Render, run, and lint prompts directly from your terminal for rapid development and testing.
• 🧪 Fully Tested & Typed: A comprehensive test suite and full type-hinting ensure reliability.

🚀 Quick Start

1. Installation

pip install promptkit-core

2. Define a Prompt

Create a file named prompts/generate_greeting.yaml:

# prompts/generate_greeting.yaml
name: generate_greeting
description: "Generates a personalized and professional greeting."
template: |
  Hello {{ name }},

  Welcome to the team! We are excited to have a {{ role }} with your skills on board.

  Best,
  The PromptKit Team
input_schema:
  name: str
  role: str

3. Use in Python

# main.py
from promptkit.core.loader import load_prompt
from promptkit.core.runner import run_prompt
from promptkit.engines.openai import OpenAIEngine

# Load prompt from YAML (assuming it's in a 'prompts' directory)
prompt = load_prompt("generate_greeting", prompt_dir="prompts")

# Configure engine
engine = OpenAIEngine(api_key="sk-...") # Or load from environment

# Run prompt with validated inputs
response = run_prompt(prompt, {"name": "Alice", "role": "Software Engineer"}, engine)
print(response)

4. Use the CLI

The CLI is perfect for quick tests, rendering, and validation.

# Set the directory where your prompts are stored (optional, can be passed as an argument)
export PROMPTKIT_PROMPT_DIR=./prompts

# Run the prompt directly from the terminal
promptkit run generate_greeting --key "sk-..." --name "Bob" --role "Data Scientist"

# Just render the template to see the output
promptkit render generate_greeting --name "Charlie" --role "Product Manager"

# Lint your YAML file to check for errors
promptkit lint generate_greeting

📚 Documentation

For detailed usage, advanced features, and API reference, please refer to the Official PromptKit Documentation.

🤝 Contributing

Contributions are welcome! Whether it's a bug report, a new feature, or a documentation improvement, please feel free to open an issue or submit a pull request.

1. Fork the repository.
2. Create your feature branch (git checkout -b feature/AmazingFeature).
3. Install development dependencies: pip install -e '.[dev]'
4. Commit your changes (git commit -m 'Add some AmazingFeature').
5. Push to the branch (git push origin feature/AmazingFeature).
6. Open a Pull Request.

Please see the CONTRIBUTING.md file for more details.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.