Ariamehr's Gemini Pandas Agent

A Google-native alternative to LangChain's create_pandas_dataframe_agent, powered by Gemini and LangGraph.

Created by Ariamehr A

Installation

pip install ariamehr-gemini-pandas-agent

Quick Start

import pandas as pd
import google.generativeai as genai
from ariamehr_gpandas_agent import create_pandas_dataframe_agent

# Configure Gemini
genai.configure(api_key="YOUR_API_KEY")
model = genai.GenerativeModel("gemini-2.0-pro")

# Load your data
df = pd.read_csv("your_data.csv")

# Create the agent
agent = create_pandas_dataframe_agent(model, df, allow_dangerous_code=False)

# Ask questions about your data
response = agent.run("What is the average value in column X?")
print(response)

Features

• 🚀 Powered by Google's Gemini API and LangGraph
• 🔒 Secure Python AST-based code execution
• 📊 Support for single or multiple DataFrames
• 🎯 Simple, LangChain-compatible interface
• 🛡️ Safe by default with optional dangerous code allowance

Security Note

By default, the agent runs in a secure mode that prevents potentially dangerous imports. If you need to allow specific imports, use allow_dangerous_code=True, but be aware of the security implications.

License

MIT License - see LICENSE file for details.

Author

Created by Ariamehr A

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.