sandbox-execution-fallback

SKILL.md

---
name: sandbox-execution-fallback
description: Recover from execute_code_sandbox failures by writing Python scripts to files and executing via run_shell
---

# Sandbox Execution Fallback

## When to Use

Apply this pattern when `execute_code_sandbox` fails or times out, particularly for:

- Spreadsheet generation (pandas, openpyxl, xlsxwriter)
- Complex data processing with file I/O
- Tasks requiring external library imports
- Long-running computations that hit timeout limits

## Recovery Procedure

### Step 1: Capture the Python Code
Extract or reconstruct the Python code that failed in `execute_code_sandbox`.

### Step 2: Write Code to File
Use `write_file` to save the script with a `.py` extension:

```
write_file(
    path="script_name.py",
    content="""
import pandas as pd
# Your implementation here
"""
)
```

### Step 3: Execute via run_shell
Run the script using the system Python interpreter:

```
run_shell(command="python3 script_name.py")
```

### Step 4: Verify Output
Confirm the results match expected outputs (files created, data processed correctly, etc.).

## Complete Example

```
# Failed: execute_code_sandbox with pandas Excel generation

# Recovery - Step 1 & 2: Write script to file
write_file(
    path="generate_pnl_report.py",
    content="""
import pandas as pd
from openpyxl import Workbook

# Create sample data
data = {
    'Category': ['Revenue', 'Expenses', 'Tax'],
    'Amount': [10000, 3000, 500]
}
df = pd.DataFrame(data)

# Write to Excel
df.to_excel('pnl_report.xlsx', index=False)
print('Report generated: pnl_report.xlsx')
"""
)

# Step 3: Execute via shell
run_shell(command="python3 generate_pnl_report.py")

# Step 4: Verify file was created
run_shell(command="ls -la pnl_report.xlsx")
```

## Why This Works

| Aspect | execute_code_sandbox | run_shell + write_file |
|--------|---------------------|------------------------|
| Environment | Sandboxed, limited | Full system Python |
| File I/O | Restricted | Full access |
| Timeout | Strict limits | More flexible |
| Library Support | May be limited | System-installed packages |
| Result | Identical output | Identical output |

## Best Practices

1. **Use descriptive filenames** - e.g., `generate_report.py`, `process_data.py`
2. **Add error handling** - Include try/except blocks in your script
3. **Print progress** - Use print statements for debugging
4. **Clean up** - Remove temporary scripts after successful execution if needed
5. **Verify results** - Always confirm outputs before proceeding

## Common Use Cases

- Excel/CSV report generation
- Data transformation pipelines
- Batch file processing
- API data aggregation
- Chart and visualization creation

## Troubleshooting

If `run_shell` also fails:

1. Check Python is available: `run_shell(command="python3 --version")`
2. Install missing packages: `run_shell(command="pip3 install pandas openpyxl")`
3. Check file permissions and paths
4. Review stderr output for specific errors