resilient-spreadsheet-workflow

Resilient Spreadsheet Workflow

This skill provides a robust workflow for processing spreadsheet files (CSV, XLSX) when execute_code_sandbox encounters failures. It uses a combination of tools to ensure reliable execution and clear error diagnosis.

When to Use

• Processing spreadsheet files when execute_code_sandbox returns unknown errors
• Need to debug why spreadsheet operations fail
• Require reliable file I/O with verification at each step

Step-by-Step Workflow

Step 1: Initial Exploration with shell_agent

Start by using shell_agent to explore the file structure and understand what files exist:

Use shell_agent to:
- List files in the working directory
- Identify spreadsheet files (CSV, XLSX)
- Examine file sizes and basic structure

Example task for shell_agent:

Explore the current directory to find all spreadsheet files (CSV, XLSX). 
List their sizes and identify which files need processing.

Step 2: Write Processing Script with write_file

Create a standalone script file rather than executing inline code:

# Use write_file to create a script like "process_sheet.py"
content = """
import pandas as pd
import sys

try:
    df = pd.read_csv('input.csv')  # or pd.read_excel for XLSX
    # Perform your processing
    result = df.describe()
    result.to_csv('output.csv', index=False)
    print("Processing complete")
except Exception as e:
    print(f"Error: {e}", file=sys.stderr)
    sys.exit(1)
"""

write_file(path="process_sheet.py", content=content)

Step 3: Execute with run_shell and Error Redirection

Execute the script with stderr redirected to stdout for complete error capture:

run_shell command="python process_sheet.py 2>&1"

The 2>&1 redirection ensures both stdout and stderr are captured together, making error messages visible.

Step 4: Isolate Errors with Verification Commands

If Step 3 fails, add targeted verification commands to diagnose the actual problem:

# Check if Python is available
run_shell command="which python 2>&1"

# Check if pandas is installed
run_shell command="python -c 'import pandas' 2>&1"

# Verify input file exists and is readable
run_shell command="ls -la input.csv 2>&1"

# Check file encoding/content
run_shell command="file input.csv 2>&1"
run_shell command="head -5 input.csv 2>&1"