Data Lineage Tracer Protocol

This skill answers the critical question: "If I change this column, who dies?" It traces the flow of data backward (Upstream) to its source, and forward (Downstream) to its consumers.

Core assumption: Data does not exist in a vacuum. It flows through Views, Materialized Views, Stored Procedures, Triggers, and ORM boundaries.

1. Upstream & Downstream Mapping (Static vs Dynamic)

• Default (Static): Analyze based on provided .sql, schema files, and application ORM code (e.g. Prisma models).
• Dynamic (On-Demand): Only connect to the database to inspect dynamic dependencies (e.g. live pg_views or information_schema.triggers) if requested explicitly. If assessing reporting.daily_sales.total_amount:
• Is it a physical physical or a computed column?
• Does a VIEW project it directly from public.orders?
• Is it populated via a daily INSERT INTO ... SELECT cron job?