What problem does it solve? Onboarding a new raw retail table into a governed analytics pipeline requires profiling the source, authoring mapping artifacts, and recording readiness status, all without skipping human judgment calls on grain, PII, and business rollups. This Skill sequences that first stage transition so no table reaches mapping without evidence and review. ## Core Features & Use Cases - Stage 1 Source Ready profiling: Measures row counts, column missingness, candidate-key uniqueness, and returns-column population over a read-only connection, recording results in source-profile.md. - Stage 2 mapping delegation: Invokes the source-mapping skill to author source-map.yaml, assumptions.md, unresolved-questions.md, and the reconciliation report blank, without duplicating the mapping procedure. - Readiness bookkeeping and hard stops: Seeds readiness-status.yaml with explicit pass/warning/blocked statuses plus evidence, and halts at four human decision seams (grain, PII publish-safety, business rollup, product identity). - Use Case: When a new table like raw store transactions arrives, ask the agent to onboard it; it profiles the data, delegates the mapping artifacts, flags open judgment questions for named owners, and stops at Mapping Ready pending human approval. ## Quick Start Ask the agent to onboard the new raw table by saying "onboard schema.table from nothing to a reviewed map" and let it profile, delegate mapping, and seed the readiness status.
