build-data-brieflisted

# Build: /data-brief You are about to generate a `/data-brief` skill customised to this company's request flow. The premise: the requested dashboard is usually a guess at a solution, and the analyst's real job is the problem behind it. This skill makes the requester state the problem before the queue accepts the request. Complete all three phases first. ## Phase 1 — Inspect 1. **Find where requests arrive today.** Slack channels referenced in READMEs or configs, Jira/Linear project keys, request forms linked from docs. Usually it is "everywhere", which is itself the finding. 2. **Find the request artefacts, if any.** Ticket templates, intake forms, a half-abandoned "data request" doc. The generated skill builds on whatever shape already has adoption rather than inventing a rival. 3. **Check the semantic layer.** dbt metrics, BI tool definitions, a metric dictionary. A brief that names a governed metric can be answered fast; a brief that invents one cannot — the skill needs to know which is which. 4. **Check the installed skills.** If `/support-triage` exists (perhaps from this foundry), the brief skill becomes its natural next step for the "data question" shape; if `/settle-the-number` exists, disputed definitions route there. ## Phase 2 — Interview 1. **"Who triages requests today, and what do they reject or bounce most?"** The top three bounce reasons become the brief's required fields — the skill asks upfront what the triager would have asked two days later. 2. **"Wha