Scans directory structure, detects projects, maps dependencies, and documents code organization into a repo.md file. Use when the user needs a codebase overview, project structure map, or repository context before security analysis.
The ghost-repo-context skill builds a repository overview by scanning a codebase's directory structure, detecting its projects, summarizing their architecture, and writing the results to a repo.md file. It is intended to produce repository context before security analysis, and does all the work locally without spawning subagents.
It accepts a repository path and an optional cache directory (computing a stable cache id from the repo name and remote URL hash when not given), and is cache-first: if a repo.md already exists it returns immediately. Otherwise it detects projects using a bundled detector, summarizes each project with a bundled summarizer (capturing architecture, sensitive data types, business criticality, and a component map), writes the combined results into repo.md using a template, validates the output, and returns its path. It is restricted to local files only, using no web fetch or search.
It is most useful as a preparatory step that gives a security-analysis or code-review workflow a structured map of what is in a repository and how it is organized.
A repo.md file documenting the repository's detected projects, their architecture summaries, sensitive data types, business criticality, and component maps.
When you need a codebase overview, project structure map, or repository context, typically before running security analysis.
Yes. It is cache-first: if repo.md already exists in the cache directory it returns that path immediately instead of re-scanning.
No. It is restricted to local files and does not use web fetch or search.
It detects projects with a bundled detector, summarizes each with a bundled summarizer, writes a templated repo.md, and validates the result before returning its path.
Quick Setup:
.claude/skills/Repository
ghostsecurity/skills