maestro:parallel-exploration

Parallel code exploration works by running concurrent read-only scouts across multiple domains simultaneously. This approach saves significant time compared to sequential investigation when analyzing how features are implemented across different parts of a large codebase.

Investigating code structure and test cases simultaneously requires defining independent sub-questions and crafting specific scout prompts. The parallel exploration approach executes these concurrently to gather targeted evidence about code structure, data flow, and configuration settings.

Parallel investigation for code exploration should be used when you need a deep understanding of a subsystem before implementation. It applies to scenarios requiring concurrent task execution across multiple domains, such as understanding a new feature implementation across a large codebase.

Parallel code exploration does not modify files during investigation as it operates strictly as a read-only process. It gathers evidence about code structure, data flow, and configuration points without making changes to the codebase.