researching-codebase

Codebase exploration involves translating a vague research objective into a structured plan, using a file-finder to locate relevant files, reading core logic, and documenting findings to understand the system before planning implementation.

The best way to research and document a codebase is through a phase-driven workflow: understand the objective, explore files with a file-finder, document findings in a research report, and transition to implementation planning with validation questions after each phase.

You trace data flow and locate relevant files during codebase research by executing a structured exploration plan that uses a file-finder tool to identify core files, reads their contents, and documents the findings in a staged research report.

Yes, you can trigger web research for external libraries during codebase exploration to gather necessary context about unfamiliar APIs or dependencies encountered while reading core files and tracing data flow.

Structured codebase research generates documentation output by creating a research plan directory containing a research.md file for findings and a plan.md file to guide subsequent implementation steps.

You should use a question-first approach for codebase exploration when you have a vague request requiring clarification, ensuring you understand the research objective before touching the code or initiating any file-finder searches.