parallel-exploration

To explore a codebase in parallel, you can decompose a complex question into independent sub-questions and investigate code, tests, and documentation concurrently using a fan-out strategy. This read-only approach gathers information without modifying files.

The best way to research multiple code questions concurrently is using a fan-out exploration strategy with a scout-researcher subagent. This allows you to simultaneously gather information across codebase, tests, documentation, and configuration domains for faster research.

Yes, you can use parallel exploration to investigate test coverage and code logic simultaneously. It operates as a read-only research mechanism designed for understanding existing systems safely during planning and exploratory phases without modifying any files.

You fan out sub-questions by breaking a complex feature investigation into smaller, parallelizable tasks. This lets you concurrently ask where core logic is implemented, how the feature is tested, and what configuration or documentation says about the pattern.

Yes, parallel codebase exploration fully supports read-only research for planning phases. It leverages a scout-researcher subagent to gather information across multiple domains concurrently, making it ideal for understanding systems before making changes.

You should not use a fan-out strategy for codebase research when you need to modify files or write code, as this approach is strictly read-only. It is designed for investigation and understanding rather than implementation or altering existing systems.