The model didn't run the script. As pointed out by @zozbot234 in another response, it would need to be run in an agentic harness. This prompt was executed in LMStudio, so just inference.