research-explore

Codebase architecture investigation works by decomposing questions into actionable sub-tasks, delegating them to an explore agent for parallel execution, and synthesizing findings into a structured, report-ready document.

Running parallel research workflows is best achieved by decomposing your question, delegating sub-tasks to an exploration agent simultaneously, and synthesizing the multi-source findings into a structured report.

Yes, you can use this approach for read-only investigation across both the codebase and external sources, coordinating end-to-end research workflows to synthesize multi-source inquiries into a structured report.

To generate a structured report from cross-file analysis, decompose your research question, delegate sub-tasks to an explore agent in parallel, and synthesize the aggregated findings into a report-ready document.

You should avoid delegating sub-tasks to an explore agent for research synthesis when your inquiry is simple enough to answer directly, as this approach requires a delegation mechanism and parallel task orchestration designed for complex, multi-source investigations.