repo-research

To generate a project context document, this Skill scans the repository structure and surface-level assets, writing the findings into `.agents/rules/project-context.md`. It builds a clear project context progressively through phased scanning rather than bulk-reading source files.

Phased repository analysis is a structured discovery method that starts with a top-level repo map, then progressively reveals deeper insights like config files, entry points, and data models. This approach avoids bulk-reading source files and incrementally appends findings to the project context document.

Yes, you can analyze a large codebase without bulk-reading by using the phased scanning approach. It limits the initial findings to a top-level repo map and progressively reveals deeper insights, appending updates per phase to avoid overwhelming context windows.

To document repository structure and entry points, the Skill identifies root config files and data models, then precisely records these findings into `.agents/rules/project-context.md`. This creates a living project context document for ongoing software research.

No specific dependencies or components are required to run this repository analysis. It operates independently to scan the repo structure and generate the project context document without external tool integrations.

The limitation of phased scanning for repo research is that it avoids bulk-reading source files, relying instead on progressively revealing surface-level assets and structural insights. This means it focuses on high-level architecture and config rather than deep line-by-line logic.