code_interpreter

Overview

code_interpreter executes arbitrary Python code and returns stdout / stderr plus generated artifacts.

It is the main tool for data analysis, calculations, dataframe operations, and chart generation.

Parameters

{
  "code": "python code string"
}

What it does

• runs Python code in a subprocess
• provides common analysis packages such as pandas and numpy
• captures text output and generated images
• keeps generated image references available for later HTML rendering

When to use it

• CSV / Excel / dataframe analysis
• metrics computation
• chart generation
• Python-based preprocessing and transformation

Example

import pandas as pd

df = pd.read_csv(FILE_PATH)
print(df.head())
print(df.describe())

Notes

• each call is independent
• variables do not persist between calls
• always load the data you need inside the same call
• use print() if you want the result to appear in the tool output
• do not use it as the final HTML delivery step