A developer-friendly command-line tool for managing, testing, and comparing prompt templates with OpenAI's GPT models—designed to make prompt engineering reproducible, versionable, and fast.

🚀 Features

• Run prompt templates with dynamic input
• Compare multiple prompt versions side-by-side
• Track token usage and estimate API cost
• Log results to .txt and .csv files
• Scaffold new prompts using a consistent YAML format

🛠️ Installation

Clone the repo

git clone https://github.com/your-username/prompt-playground-cli.git cd prompt-playground-cli

Set up virtual environment

python -m venv venv source venv/Scripts/activate # On Windows

Install dependencies

pip install -r requirements.txt

Install CLI in editable mode

pip install -e .

🧪 Usage Examples

Create a new prompt scaffold

prompt-cli new summarize-v1

Example prompt file (YAML)

name: summarize-v1 model: gpt-3.5-turbo temperature: 0.7 prompt: | Summarize this in one sentence: {input}

Run a prompt

prompt-cli run summarize-v1.yaml --input "AI is transforming healthcare."

Compare multiple prompts

prompt-cli compare --input "Summarize AI trends in 2024." summarize-v1.yaml summarize-v2.yaml

Saves .txt and .csv logs to .prompt-history/

List available prompts

prompt-cli list

🔐 Environment Variables

Create a .env file in your project root:

OPENAI_API_KEY=your-openai-api-key

Or set it in your terminal session:

$env:OPENAI_API_KEY="your-openai-api-key" # Windows PowerShell

📁 File Structure

prompt-playground-cli/ ├── prompt_cli/ # CLI source code ├── prompts/ # Prompt YAML templates ├── .prompt-history/ # Output logs and comparison results ├── requirements.txt ├── setup.cfg ├── README.md └── .gitignore

📄 License

MIT License