researching-codebases

To map codebase architecture, the Skill analyzes source code to produce structured findings with evidence and file references. It decomposes research questions into focused areas to locate implementations and trace data flow paths.

Gathering technical context involves analyzing the codebase to describe what exists, where it exists, and how pieces connect. The Skill synthesizes a structured answer with targeted searches to support planning changes.

Yes, you can persist codebase research findings. The Skill supports optional persistence to docs/research/YYYY-MM-DD-<topic>.md, saving the structured architecture and data flow analysis directly as a markdown document.

Evidence-based codebase research works by using search and read operations to locate implementations, extracting concrete evidence from code and docs. This ensures architecture mapping is grounded in actual file references rather than assumptions.

No specific dependencies are required to trace call paths. The Skill operates independently using built-in file finding, reading, and searching capabilities to gather evidence and map connections across the codebase.

Automated codebase research should be avoided when you need real-time runtime analysis or execution tracing, as this Skill relies on static analysis of code and documentation files rather than dynamic program monitoring.