Search for a command to run...
Thinking to Recall: How Reasoning Unlocks Parametric Knowledge in LLMs