> ## Documentation Index > Fetch the complete documentation index at: https://docs.zerotwo.ai/llms.txt > Use this file to discover all available pages before exploring further. # Code Interpreter > Execute Python code in a secure sandbox for data analysis, calculations, and problem-solving ## What is Code Interpreter? Code Interpreter is a powerful tool that allows AI to write and execute Python code in a secure sandbox environment. It enables complex calculations, data analysis, file manipulation, and more - all within your conversation. Think of Code Interpreter as giving AI the ability to "show its work" by writing and running actual code to solve problems. ## When to Use Code Interpreter Analyze CSV files, calculate statistics, identify trends Complex calculations, equations, mathematical modeling Convert formats, extract data, manipulate files Create charts, graphs, and data visualizations Algorithm implementation, logic verification Run simulations, test scenarios, model systems ## Enabling Code Interpreter Enable for specific tasks: 1. Click **Code Interpreter** toggle in top bar 2. Toggle turns blue when active 3. Send your request 4. AI will use code when beneficial AI automatically decides when to use code based on your request. Enable by default: 1. Go to **Settings** > **Preferences** 2. Find **Code Interpreter** section 3. Toggle **Enable by default** 4. Save changes Can still toggle off for specific messages. Enable for entire project: 1. Open Project settings 2. Go to **Tools** tab 3. Enable **Code Interpreter** 4. All chats in project can use it ## How It Works AI determines if code execution would help answer your question AI writes Python code to solve the problem ```python theme={null} import pandas as pd import matplotlib.pyplot as plt # AI writes relevant code data = pd.read_csv('sales_data.csv') total_sales = data['revenue'].sum() ``` Code runs in secure sandbox environment: * Isolated from your system * No internet access * Time-limited execution * Safe and contained AI receives output and incorporates it into response: * Calculation results * Generated visualizations * Error messages (if any) * Data outputs AI explains the code and results in natural language ## What You Can Do ### Data Analysis **Analyze uploaded CSV files**: ``` Analyze this sales data and tell me: 1. Total revenue 2. Best performing month 3. Average order value 4. Revenue trend over time [upload sales_data.csv] ``` AI will: * Read and parse the CSV * Calculate requested metrics * Identify patterns * Create visualizations **Calculate statistics**: ``` Calculate mean, median, mode, and standard deviation for the price column in this dataset ``` ```python theme={null} # AI-generated code stats = { 'mean': data['price'].mean(), 'median': data['price'].median(), 'mode': data['price'].mode()[0], 'std': data['price'].std() } ``` **Clean and prepare data**: ``` Clean this dataset: - Remove duplicates - Handle missing values - Standardize date format - Remove outliers ``` AI writes code to clean and export cleaned data. **Identify patterns**: ``` Plot the trend of website visits over the past 6 months and identify any seasonal patterns ``` Creates visualizations and analysis. ### Mathematical Operations **Solve equations**: ``` Calculate compound interest on $10,000 invested at 5% annual rate, compounded quarterly for 10 years ``` ```python theme={null} principal = 10000 rate = 0.05 times_compounded = 4 years = 10 amount = principal * (1 + rate/times_compounded)**(times_compounded*years) interest = amount - principal ``` **Linear algebra**: ``` Multiply these two matrices and find the determinant Matrix A: [[1,2],[3,4]] Matrix B: [[5,6],[7,8]] ``` Uses NumPy for matrix operations. **Finance and accounting**: ``` Calculate NPV for a project with initial investment of $100k and cash flows of $30k, $40k, $50k over 3 years at 10% discount rate ``` Shows full calculation with explanation. ### Visualizations **Create visualizations**: ``` Create a bar chart comparing revenue across regions ``` ```python theme={null} import matplotlib.pyplot as plt regions = ['North', 'South', 'East', 'West'] revenue = [45000, 38000, 52000, 41000] plt.bar(regions, revenue) plt.title('Revenue by Region') plt.xlabel('Region') plt.ylabel('Revenue ($)') plt.show() ``` Chart appears directly in chat. **Line plots, scatter plots**: ``` Plot the relationship between advertising spend and sales revenue ``` Creates scatter plot with trend line. **Correlation analysis**: ``` Create a heatmap showing correlations between all numeric columns in the dataset ``` Visual correlation matrix. ### File Operations **Convert between formats**: ``` Convert this CSV to JSON format ``` ``` Convert this JSON to a formatted Excel file ``` Downloads converted file. **Extract specific data**: ``` Extract all email addresses from this text file ``` ``` Pull out dates and amounts from these receipts ``` Returns structured data. **Create files**: ``` Generate a CSV file with 100 sample customer records including name, email, and purchase history ``` Creates and provides download link. ## Understanding Code Output ### Reading Results AI shows the code it's executing: ```python theme={null} # Calculating total sales df = pd.read_csv('sales.csv') total = df['amount'].sum() print(f"Total sales: ${total:,.2f}") ``` * Syntax highlighted * Comments explaining steps * Copyable with one click Results appear in output block: ``` Total sales: $1,234,567.89 Top 5 products: 1. Widget A - $450,233 2. Gadget B - $389,012 3. Tool C - $278,901 4. Device D - $234,567 5. Gizmo E - $198,432 ``` Clear, formatted output. Charts and graphs display inline: !\[Example chart showing data visualization] * High-quality images * Downloadable * Regenerate with different styles If code fails: ``` Error: KeyError: 'revenue' Column 'revenue' not found in dataframe. Available columns: ['date', 'product', 'sales_amount'] ``` * Clear error description * What went wrong * AI suggests fix ## Available Python Libraries Code Interpreter includes many popular libraries: ### Data Analysis ```python theme={null} import pandas as pd # Data manipulation import numpy as np # Numerical computing from scipy import stats # Statistical functions ``` ### Visualization ```python theme={null} import matplotlib.pyplot as plt # Plotting import seaborn as sns # Statistical viz ``` ### Machine Learning ```python theme={null} from sklearn import * # Scikit-learn ``` ### Utilities ```python theme={null} import json # JSON handling import csv # CSV operations import datetime # Date/time import re # Regular expressions import math # Mathematical functions ``` Most popular Python libraries are available. If you need something specific, just ask! ## Advanced Features ### Multi-Step Operations Complex workflows with multiple code executions: ```python theme={null} # Load data df = pd.read_csv('data.csv') # Clean data df = df.dropna() df['date'] = pd.to_datetime(df['date']) ``` ```python theme={null} # Calculate metrics df['month'] = df['date'].dt.month monthly_sales = df.groupby('month')['revenue'].sum() ``` ```python theme={null} # Create chart plt.plot(monthly_sales.index, monthly_sales.values) plt.title('Monthly Revenue Trend') plt.show() ``` ```python theme={null} # Save results monthly_sales.to_csv('monthly_summary.csv') ``` ### Interactive Analysis Have conversations about the code: ``` You: Analyze this dataset for trends AI: [Runs analysis code] Here's what I found... You: Can you break down the results by region? AI: [Modifies code to add regional breakdown] Sure, here's the regional analysis... You: Now create a visualization AI: [Generates chart code] Here's the visualization... ``` ### Error Recovery AI handles errors gracefully: 1. **Code fails** → AI sees error 2. **AI analyzes** → Understands problem 3. **AI fixes** → Rewrites code 4. **Code runs** → Success! You see the fix and explanation. ## Security & Limitations ### Sandbox Environment Code runs completely isolated from your system Cannot make external network requests Execution timeout prevents infinite loops Limited RAM prevents resource exhaustion ### What's Allowed ✅ File operations (uploaded files only)\ ✅ Data analysis and calculations\ ✅ Image generation (charts/graphs)\ ✅ Mathematical operations\ ✅ String manipulation\ ✅ Data transformations ### What's Not Allowed ❌ Network requests\ ❌ System file access\ ❌ Installing packages\ ❌ Long-running operations (>60s)\ ❌ Large file generation (>50MB)\ ❌ Cryptocurrency mining Code Interpreter is powerful but sandboxed for security. It cannot access your local files or make network requests. ## Best Practices **Better requests**: ❌ "Analyze this data"\ ✅ "Calculate the average, median, and identify the top 10 highest values in the 'sales' column" Specificity helps AI write targeted code. **Include relevant information**: ``` This CSV contains sales data with columns: - date (YYYY-MM-DD format) - product_id - quantity - price - region Calculate total revenue by region ``` Context prevents errors. **Build complexity gradually**: 1. First: "Load and show first 5 rows" 2. Then: "Calculate summary statistics" 3. Finally: "Create detailed visualization" Step-by-step is more reliable. **Check important calculations**: * Spot check numbers * Verify logic makes sense * Compare with expected ranges * Ask AI to explain unusual results AI is powerful but can make mistakes. ## Example Use Cases ### Sales Analysis ``` Analyze this sales data: 1. Total revenue by month 2. Best performing products 3. Growth rate month-over-month 4. Create a line chart of revenue trend 5. Identify any anomalies [upload sales_data.csv] ``` ### Scientific Calculations ``` Calculate the trajectory of a projectile launched at: - Initial velocity: 50 m/s - Angle: 45 degrees - Height: 10 meters Show: - Maximum height reached - Total distance traveled - Time of flight - Plot the trajectory ``` ### Data Transformation ``` Transform this customer data: 1. Convert dates to ISO format 2. Standardize phone numbers 3. Clean up email addresses 4. Remove duplicates 5. Export as clean CSV ``` ## Troubleshooting **Possible causes**: * Code Interpreter not enabled * Syntax error in code * Execution timeout **Solutions**: * Verify toggle is on * Ask AI to fix errors * Simplify the operation **Common issues**: * Misunderstood requirements * Wrong assumptions about data * Logic error in code **Solutions**: * Clarify your requirements * Provide data structure details * Ask AI to explain its approach * Verify step by step **Check**: * File was uploaded successfully * Correct filename used * File format supported **Solutions**: * Re-upload file * Check filename spelling * Convert to supported format ## Pricing & Limits ### Free Tier * 10 code executions per day * 30 second timeout * Standard Python libraries * 10MB max file size ### Pro Tier * 100 code executions per day * 60 second timeout * All Python libraries * 50MB max file size ### Team Tier * Unlimited executions * 120 second timeout * Custom library requests * 100MB max file size Execution limits reset daily at midnight UTC. ## Next Steps Combine with research for data-driven insights Edit code in Code Canvas for larger projects Learn about file upload capabilities Combine with real-time data retrieval Code Interpreter transforms AI from answering questions to solving problems with actual computation!