acquire-codebase-knowledge

To generate codebase documentation for onboarding, you scan an unfamiliar repository to produce seven structured, evidence-backed documents covering stack, structure, architecture, conventions, integrations, testing, and concerns. Each document maps the codebase using verifiable facts traced directly from source files.

Evidence-based codebase documentation strictly grounds every architectural claim in existing project files or terminal output. It avoids unsupported inference by marking unknown details as [TODO] and flagging intent-dependent decisions as [ASK USER], ensuring the architecture mapping remains verifiable.

Yes, you can provide a specific focus area to prioritize generating matching onboarding documents within the large codebase. The system will prioritize those specific documents while still running a final validation pass across all seven required documentation areas.

You map software architecture without guessing by scanning project files for intent documents and emitting exactly seven structured docs under docs/codebase. Unknowns are marked as [TODO] and intent gaps as [ASK USER], ensuring conventions and integrations are backed by concrete evidence lists.

The limitation of automated repository analysis is that it never infers unverifiable details. If testing documentation or integration patterns lack source evidence, the onboarding docs will explicitly mark these gaps as [TODO] or [ASK USER] rather than generating speculative explanations.