← ClaudeAtlas

docs-bootstraplisted

Bootstrap or extend layered, machine-checked documentation for a codebase — research that separates what was verified from what was assumed, features with BDD scenarios, client screens, API endpoint references, service documents and a file-per-item backlog, all carrying paths into the code. Use when a repository has no documentation an agent can navigate, when documentation exists but has drifted from the code, or when a new feature needs documenting before it is built.
youndie/docs-bootstrap · ★ 2 · Data & Documents · score 68
Install: claude install-skill youndie/docs-bootstrap
# docs-bootstrap Produce documentation whose primary consumer is a coding agent: layers linked by ids, every document carrying paths into the code, every claim checkable by machine. ``` [ Research — why the architecture is this and not that; verified vs hypothesis ] optional │ [ Feature — what the system does and why, + BDD scenarios = acceptance criteria ] │ [ Screen / flow — what the user sees ] only if there is a client │ [ API — the contract: URL, auth tier, error codes ] │ [ Service — who owns the data, config, deploy, quirks ] ``` Each layer answers its own question and points at its neighbour by id. Read [SPEC.md](SPEC.md) before writing anything — it is the contract. [example/](example/) is a complete worked instance of it; when unsure what a filled-in document should look like, read the example rather than guessing. ## The two rules everything else follows from > **`main` describes what exists. An open pull request describes what will be.** Never document intent as fact. A feature that is designed but not shipped is `status: draft` and lives in an open pull request. What is not built yet is either that, or labelled *target* or *hypothesis* in the text itself. > **What was verified is separated from what was assumed, explicitly.** Every status code, error string, field name and limit must be read out of the source before it is wr