@[Andrey Turkin]
Great question! It actually does both — and that's what makes it different from typical code-to-docs tools.
When a commit hits your repo, RepoWrit doesn't just look at the diff in isolation. It pulls in the full file tree, recent commit history, existing README/CHANGELOG, and analyzes module relationships to understand why a change was made, not just what changed. So if you refactor an auth module, it knows that impacts your middleware, API routes, and session handling — and writes documentation that reflects that context.
For brand new repos, it takes a full snapshot — file structure, code samples across the codebase — and generates docs that capture the actual architecture, not just a list of files.
TL;DR: it understands project context deeply. The "autopilot" part is real — it hooks into your GitHub pushes so docs stay current without you thinking about it. Happy to show you a demo if you're curious!
