add-repositorylisted
Install: claude install-skill kvdm-co-pilot/create-cmp
# add-repository — stamp a data/domain slice only
> Deterministic-stamp, gate-proven. The script (`qa/scaffold-feature.mjs --preset repository`)
> does the mechanical work — copy the exemplar's data/domain files, whole-word identifier
> rename, DI injection for the repository + use case only. You (the AI) only customize the
> entity's fields and its sample data. You are not done until `node qa/verify.mjs` PASSes and the
> receipt is committed — see this project's `CLAUDE.md`.
This is the `repository` subset of `add-feature` — same stamper, same rename mechanic, filtered
to just the 5 data/domain files. Use it when you want the data layer to exist before any screen
consumes it (e.g. you're modeling several entities up front, or a screen for this entity will
come later via `add-screen`).
**Name the lane first, out loud.** Before any tool runs, tell the human in a sentence which
lane this request takes (`CLAUDE.md` §"After genesis"): an entity carrying decisions worth
recording (data-model semantics, boundary rules) goes through a feature brief
(`docs/features/<name>.md`, signed before code); an ordinary entity is direct lane — say so,
then continue here. Never route silently.
## Why a stamper and not hand-written files
Every hand-written file is a drift chance in this project's architecture. `qa/scaffold-feature.mjs
--preset repository` produces a **conforming skeleton by construction** for the data/domain
layer: a domain model, a domain repository interface, a use case, a