atomic-hld-design

What problem does it solve?

This Skill eliminates inconsistent, non-compliant High-Level Design (HLD) documentation for Atomic layer source systems in the Medallion data architecture, which causes misalignment between source data structures, business concept validation (BCV) mappings, and downstream low-level design (LLD) and ETL development for the UBCK Lakehouse project. It removes manual guesswork from tiered entity design, table type classification, scope categorization, and ensures adherence to project-wide standards to reduce review cycles and metadata errors.

Core Features & Use Cases

• End-to-end HLD workflow: Guided step-by-step process for designing or updating HLD for new or existing source systems (DCST, FMS, NHNCK, etc.), including input collection from BRD files, table type classification, tiered dependency mapping, BCV term lookup, shared entity review, and out-of-scope table categorization.
• Standardized output enforcement: Strict formatting and content rules for HLD Tier and Overview files to ensure compatibility with automated aggregation scripts that generate project-wide metadata files (atomic_entities.yaml, atomic_out_of_scope.yaml), reducing manual errors and review cycles.
• Built-in reference materials: Includes standard classification group lists, file layout rules, and correct/incorrect format examples for out-of-scope tables to help data modelers avoid common mistakes.
• Use Case: A data modeler tasked with designing HLD for a new FMS source system can follow the skill's process to correctly classify tables, map BCV terms, generate compliant Tier and Overview files, and automatically update project metadata without manual formatting errors.

Quick Start

Use the atomic-hld-design skill to create a compliant HLD Tier 1 file for the new DCST source system, following the provided templates and BCV lookup steps to map source tables to Atomic entities and classify out-of-scope tables correctly.