cnb-ds-edalisted

# CNB DS EDA Use this skill to produce analysis that a stakeholder can read directly in GitHub and that another analyst can refresh from the warehouse. ## Workflow 1. Start in `/Users/vanducng/git/work/cnb/cnb-ds-eda` unless the user gives another checkout. 2. Inspect existing notebook patterns before editing. Prefer `notebooks/rnd_disconnection_status/`, `notebooks/lead_age_performance/`, and related topic folders over inventing a new structure. 3. Translate relative date requests into exact `America/New_York` report boundaries. State `start_date`, `end_exclusive`, and the inclusive last data date in the notebook. 4. Put runnable SQL in `notebooks/<topic>/queries/`. Put committed offline snapshot tables in `notebooks/<topic>/snapshots/` when GitHub rendering or offline review matters. 5. Use `src/connectors/snowflake.py::query_snowflake()` for live refresh paths. Use `vd:miudb` only when the user explicitly asks for that workflow or when the connector is insufficient. 6. When the user mentions data contracts, inspect the source YAML under `/Users/vanducng/git/work/cnb/cnb-data-contract/contracts/constraints/snowflake/` and mirror the business predicates in the analysis SQL. 7. Keep notebook code cells compact. Do not embed large TSV/CSV payloads inside code cells; load snapshot files instead. 8. Execute the notebook in-place before sharing or pushing when the user wants GitHub preview. Commit rendered outputs and PNG-backed Plotly figures. 9. Validate before handoff: `jq