Why hard code in source code of VertexAiCodeExecutor?

[NBestsuNgan]

In data-science-agent from this video data-science if you looking in the prompt of Analytic subsagent, there will be unknow name like explore_df which define in _IMPORTED_LIBRARIES of source code and we can't edit it, so that mean in our custom use case like a want to add my own function. I cannot do it right?

the sample is really great but only work with only usecase in video, VertexAiCodeExecutor have many more library out of this like CV2, seaborn, etc. but we cannot use it because we cannot edit this default value in source code.

And the person who wonder why didn't I just edit it, because I will deploy it to vertex ai agent engine and plugins into agentspace so edit the source code can't be do for my case.

please give me suggestion how can I adjust it or another way because the things that I only need is just show a custom image that all.

_IMPORTED_LIBRARIES = '''
import io
import math
import re

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import scipy

def crop(s: str, max_chars: int = 64) -> str:
  """Crops a string to max_chars characters."""
  return s[: max_chars - 3] + '...' if len(s) > max_chars else s


def explore_df(df: pd.DataFrame) -> None:
  """Prints some information about a pandas DataFrame."""

  with pd.option_context(
      'display.max_columns', None, 'display.expand_frame_repr', False
  ):
    # Print the column names to never encounter KeyError when selecting one.
    df_dtypes = df.dtypes

    # Obtain information about data types and missing values.
    df_nulls = (len(df) - df.isnull().sum()).apply(
        lambda x: f'{x} / {df.shape[0]} non-null'
    )

    # Explore unique total values in columns using `.unique()`.
    df_unique_count = df.apply(lambda x: len(x.unique()))

    # Explore unique values in columns using `.unique()`.
    df_unique = df.apply(lambda x: crop(str(list(x.unique()))))

    df_info = pd.concat(
        (
            df_dtypes.rename('Dtype'),
            df_nulls.rename('Non-Null Count'),
            df_unique_count.rename('Unique Values Count'),
            df_unique.rename('Unique Values'),
        ),
        axis=1,
    )
    df_info.index.name = 'Columns'
    print(f"""Total rows: {df.shape[0]}
Total columns: {df.shape[1]}

{df_info}""")
'''