Repository Context Builder

The problem

Entering an unfamiliar codebase often leads to hours of aimless file-hopping and architectural misunderstandings. Manual discovery misses hidden risk zones, undocumented conventions, and critical entry points, resulting in broken builds or regression errors.

What it does

• Generates an evidence-based map of repository architecture and module boundaries.
• Identifies critical entry points for APIs, workers, and application startup.
• Extracts naming, error-handling, and testing conventions from existing code patterns.
• Documents essential build, lint, and test commands without executing untrusted code.
• Highlights high-risk areas like authentication, billing, and data migrations.

Why this beats prompting it yourself

Generic LLM prompts often hallucinate architecture or get lost in node_modules. This skill follows a rigorous multi-step rubric to separate verified facts from inferences and ignores noise like build artifacts or vendored dependencies to keep context windows focused.

Use cases

• Onboarding to a new project or legacy codebase.
• Preparing context for a major refactor or feature migration.
• Creating CLAUDE.md or Cursor rules tailored to a specific repo.
• Conducting a handoff for other developers or AI coding agents.

Known limitations

Does not execute code or verify if build commands actually work. Analysis is limited to one repository per run.