Sandbox Fallback Execution

When execute_code_sandbox repeatedly fails with e2b sandbox errors, use this fallback pattern to execute Python code directly on the host system.

When to Use

Use this skill when:

• execute_code_sandbox fails with e2b-related errors
• Multiple retry attempts (3+) have not resolved the issue
• The code does not require sandbox isolation

Steps

Step 1: Write Python Script to File

Use write_file to save your Python code as a .py file:

Tool: write_file
path: /workspace/script_name.py
content: |
  # Your Python code here
  import pandas as pd
  # ... rest of code

Step 2: Execute via Shell

Use run_shell with explicit python3 path:

Tool: run_shell
command: python3 /workspace/script_name.py
timeout: 120

Step 3: Verify Output

Check the stdout/stderr from run_shell for:

• Successful execution messages
• Expected output or file creation confirmations
• Any errors that need debugging

Complete Example

# Instead of this (failing):
Tool: execute_code_sandbox
code: |
  import pandas as pd
  df = pd.DataFrame({'a': [1,2,3]})
  df.to_excel('output.xlsx')

# Use this fallback:
Tool: write_file
path: /workspace/generate_report.py
content: |
  import pandas as pd
  
  # Create sample data
  df = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6]})
  
  # Save to Excel
  df.to_excel('/workspace/output.xlsx', index=False)
  print('File created successfully')

Tool: run_shell
command: python3 /workspace/generate_report.py
timeout: 120

Tips

1. Use absolute paths in your Python script when reading/writing files
2. Add print statements for debugging since you won't get interactive output
3. Increase timeout if the script performs heavy computation
4. Check working directory - scripts run from the workspace root by default
5. Handle errors explicitly with try/except blocks and meaningful error messages

Common e2b Errors That Trigger This Fallback

• E2B execution error
• Sandbox timeout
• Connection refused
• Sandbox initialization failed
• Repeated retry failures (3+ attempts)

When NOT to Use This Fallback

• When sandbox isolation is required for security
• When the code needs specific sandbox environment variables
• On first failure only - try execute_code_sandbox retry first