get-wrapped

Turn your structured data into a yearly Wrapped-style summary, just like your favorite music recaps — but for any dataset. Works seamlessly with both OpenAI and Anthropic models.

Features

• Generate engaging, punchy recaps from any structured data.
• Provider-agnostic: works with OpenAI or Anthropic automatically.
• Simple, library-first API: just pass a summary dictionary.
• Easy integration with pandas or any analytics pipeline.

Installation

Install from PyPI:

pip install get-wrapped

Environment Variables

Set your model credentials (Not needed if you pass in your model credentials):

Anthropic Example:

export MODEL_API_URL="https://api.anthropic.com/v1/complete"
export MODEL_API_KEY="sk-xxxx"
export MODEL_NAME="claude-sonnet-4"      # optional

Or OpenAI Example:

export MODEL_API_URL="https://api.openai.com/v1/chat/completions"
export MODEL_API_KEY="sk-xxxx"
export MODEL_NAME="gpt-4.1-mini"  # optional

Usage

from get_wrapped import generate_wrapped

summary = {
    "rows": 5,
    "columns": {
        "activity": {"top_values": {"run": 3, "bike": 2}, "dtype": "object", "nulls": 0},
        "minutes": {"min": 20, "max": 60, "mean": 35, "sum": 175, "dtype": "float64", "nulls": 0},
    },
}

# Example summary (or pandas dataframe)
# summary = pd.DataFrame({
#     "activity": ["run", "run", "run", "bike", "bike"],
#     "minutes": [20, 30, 35, 30, 60]
# })

wrapped_text = generate_wrapped(summary)
print(wrapped_text)

Output -

# 🏃‍♀️ Your Activity Recap

## **Total Sessions** 📊
5 activities logged

## **Time Investment** ⏱️
175 total minutes of activity
Average session: 35 minutes

## **Activity Breakdown** 🔥
**Running** dominated with 3 sessions
**Biking** rounded out with 2 sessions

## **Session Range** 📈
Shortest: 20 minutes
Longest: 60 minutes

## **The Verdict** ✨
You kept it consistent with a solid mix of cardio activities! 🎯

How it works

1. Provide a structured summary of your dataset.
2. The library generates a prompt and sends it to your LLM (OpenAI or Anthropic).
3. Receive a fun, readable Wrapped-style recap strictly based on your data.
4. Bring your own API key—or let the library pick it up from your local environment.

Development

Clone the repository:

git clone https://github.com/kkumar/get-wrapped.git
cd get-wrapped
uv sync
uv run python examples/run_wrapped.py

License

MIT License — see LICENSE for details.