Use Claude Code for Python in VS Code on Mac

You’ve set up Python in VS Code and can write code manually. Now you want AI to help write, improve, and debug your Python code while you focus on the analysis. Think of Claude Code as a coding partner who lives inside VS Code - you describe what you want, and it writes or improves the code while you stay in the same window.

Key Concepts

Claude Code Extension - VS Code extension that brings Claude AI assistant directly into your editor with inline code edits
Chat Panel - Side panel in VS Code where you converse with Claude about your code
Inline Edits - Claude’s suggested code changes appear directly in your files with diffs you can accept or reject

What You’ll Need

Finished Python in VS Code tutorial - Python should be working in VS Code
Finished Installing Claude Code on Mac tutorial - Claude Code CLI should be installed
A paid Claude plan (such as Pro or Max) or an Anthropic Console account — the free plan doesn’t include Claude Code
20-30 minutes

Step 1: Open VS Code and Create Python Project

Open Visual Studio Code
Click File in the menu bar, then Open Folder
Navigate to your Documents folder
Click New Folder at the bottom of the dialog
Name the folder claude_python_test
Click Open to open the new folder
If prompted “Do you trust the authors?”, click Yes, I trust the authors

Click the Extensions icon in the left sidebar (four squares icon)
In the search box, type Claude Code
Find Claude Code for VS Code by Anthropic and click Install
After installation, click ✻ Claude Code in the bottom-right corner of the VS Code window (the status bar) to open the Claude panel
The first time, a sign-in screen appears: click Sign in, approve access in your browser, then return to VS Code
For detailed authentication steps, see the Installing Claude Code on Mac guide (Step 3)
Once signed in, you’ll see a Learn Claude Code checklist and a prompt box at the bottom of the panel

Step 3: Create Initial Python Script Manually

Click File > New File… in the menu bar
In the list that appears at the top of the window, choose Python File
Save the file by clicking File > Save, name it iris_analysis.py
Type this code:

from sklearn import datasets
import pandas as pd

# Load iris dataset
iris = datasets.load_iris()
iris_df = pd.DataFrame(iris.data, columns=iris.feature_names)
iris_df['species'] = iris.target_names[iris.target]

# Display structure
iris_df.info()
print("\nSummary statistics:")
print(iris_df.describe())

Save the file by clicking File > Save
Create a virtual environment, as in Step 5 of Python Coding in VS Code: click View > Command Palette, type Python: Create Environment, choose Venv, then pick your Python version
Wait for the .venv folder to appear in the Explorer
Open a terminal: click View in the menu bar, then Terminal. You should see (.venv) at the start of the prompt
Install the packages this script needs:
```
pip install scikit-learn pandas
```
In the terminal, run: python iris_analysis.py
You should see dataset structure and summary statistics in the terminal

Step 4: Ask Claude to Add Scatter Plot

Open the Claude panel (Spark icon at the top-right of the open file, or ✻ Claude Code in the bottom-right corner)
In the chat box at the bottom, type:

Add code to iris_analysis.py to create a scatter plot of sepal length vs. width, colored by species. Use seaborn.

Press Enter to send
Claude reads your file and edits it - the new code appears in iris_analysis.py
If Claude shows a diff and asks for permission, accept the change
The seaborn/matplotlib code appears in your file
Install the plotting packages Claude’s code uses (they weren’t installed in Step 3). In the terminal (with (.venv) showing), run:
```
pip install seaborn matplotlib
```

Step 5: Run the Updated Code

In the terminal, run: python iris_analysis.py
A scatter plot window appears
You should see points colored by species (setosa, versicolor, virginica)
Close the plot window to continue

Step 6: Ask Claude to Refine Plot

In the Claude chat panel, type:

Remove title. Change marker type by species. Change to the whitegrid style.

Press Enter
Claude updates the code
If asked, accept the change
Run the updated code again: python iris_analysis.py in the terminal
The plot now displays with different marker shapes per species, no title, and whitegrid style
Close the plot window to continue

Step 7: Ask Claude for PCA Analysis

In the Claude chat panel, type:

Add code to perform PCA on the numeric variables and plot the samples using the first two principal components.

Press Enter
Claude adds PCA code to your script
If asked, accept the change
Run the code: python iris_analysis.py in the terminal
A PCA plot appears showing samples projected onto PC1 and PC2, colored by species (if the scatter plot window from Step 6 opens first, close it and the PCA plot appears next)
Close the plot window when done

Step 8: Ask Claude to Review and Comment

In the Claude chat panel, type:

Review the entire script for correctness. Add comments when necessary.

Press Enter
Claude reviews the code and adds comments explaining each section
If asked, accept the change
Your script now has clear comments explaining the data loading, visualization, and PCA analysis

Next Steps

Ask Claude to create a Jupyter notebook: “Convert this to a Jupyter notebook”
Use Claude to debug errors: “This code gives Error X, can you fix it?”
Request statistical tests: “Add a t-test comparing sepal length between setosa and versicolor”
Ask Claude to optimize code: “Make this code more efficient using numpy vectorization”
Get help with unfamiliar functions: “Explain what pandas.groupby does”

Troubleshooting

Claude extension not showing - Restart VS Code after installation. Click View > Extensions to verify it’s installed.
“Cannot read Python file” error - Make sure you saved the file with .py extension. Claude needs saved files to read them.
Authentication failed - Verify your paid Claude plan or Anthropic Console account is active. See Installing Claude Code on Mac for authentication troubleshooting.
Package not found errors or command not found: python - Make sure the terminal shows (.venv) (if not, close it and open a new one with View > Terminal), then install the required packages: pip install seaborn matplotlib scikit-learn pandas. Wait for installation to complete before running code.
Code changes not applying - If Claude asked for permission, make sure you accepted the change. If it’s not working, try copying Claude’s suggested code manually.
Plot not showing - Make sure you have a display. On some systems you may need to add plt.show() at the end or use %matplotlib inline in Jupyter.

Workflow Summary

Claude Code Extension provides an AI assistant directly in VS Code with chat panel and inline diffs
Iterative refinement - Start with basic code, ask Claude to enhance, test immediately, then refine further
Context awareness - Claude reads your Python files and understands your project structure
Inline editing - Proposed changes appear as diffs in your actual files, not separate chat responses
Stay in flow - No switching between browser and editor - everything happens in VS Code

Created by Steven Ge on December 11, 2025.