← ClaudeAtlas

implementing-scriptslisted

Implement the real business logic of an already-scaffolded consumer script under scripts/<name>/ — the TDD + hub-and-spoke loop turning its contract page into tested steps/ modules. Use for "implement"/"fill in"/"wire up" a script that already has a directory. No directory yet → scaffolding-scripts first; library modules → implementing-submodules instead.
monte3l/m3l-automation · ★ 0 · Testing & QA · score 56
Install: claude install-skill monte3l/m3l-automation
# implementing-scripts This skill is the **hub playbook** for turning a scaffolded consumer script into real, reviewed, tested automation. The scaffold (from `scaffolding-scripts` / `pnpm scaffold:script`) ships a skeleton whose `steps/` bodies do nothing; the script's contract page `docs/reference/scripts/<name>.md` plus the ADR-0022 fleet conventions define what it must actually do. It is the thin counterpart to `implementing-submodules`: same operating model and spokes, but script-scale gates — there is **no coverage threshold, no exports-map/semver concern, and no provenance-count bookkeeping**; the deterministic backstop is `pnpm check:script-scaffold` and the anti-hollow gate is `pnpm knip`. ## Operating model: you are the hub, not a worker You (the main agent) **coordinate only**. You do not write `scripts/*/src` or test code yourself, and you do not review code — every substantive step runs in an isolated spoke with the right grants, so "the writer is never the reviewer" stays structural. Pass each spoke explicit context: the script name, the contract text, and concrete file paths. | Phase | Spoke (subagent) | Writes | Hand it | | --------------- | -------------------------------------------------------------------------------------------------------------------------------- | -------------