exploring-codebase

You can explore an unfamiliar codebase by running a two-phase process: a discovery pass to map architecture and boundaries, followed by adaptive deep dives that trace implementations and collect file:line evidence for actionable synthesis.

The best way to map architecture is launching autonomous explorers that reveal structural perspectives and integration points, collecting execution flow traces and cross-cutting references to synthesize targeted implementation guidance.

Finding implementation guidance and debugging starting points involves autonomously mapping repository structure and tracing execution flow to produce a structured synthesis with recommended next steps for planning and execution.

Yes, autonomous exploration applies to investigating complex bugs where architecture or integration is unclear by performing discovery passes and adaptive deep dives to collect evidence and produce debugging recommendations.

Codebase exploration works for planning cross-cutting changes by discovering natural structural perspectives, tracing cross-cutting integration points, and producing structured reports with implementation guidance and recommended next steps.

You should not use autonomous codebase exploration when you need a simple text search or when the repository is already fully understood, as its two-phase discovery and adaptive deep dives are designed for complex, unfamiliar, or multi-component codebases.