Data Architecture

When to use

Invoke after system-design has identified bounded contexts and data ownership, and before implementations/data/<engine> skills generate concrete DDL, migrations, or index definitions.

Do not use for analytics pipelines, warehouse or lakehouse modeling, stream-processing topology, single-table schema tweaks inside an already-defined data model (go directly to the engine-specific implementation skill), persistence-framework or ORM code, or isolated migration scripts.

Inputs

Required:

• Approved system-design.md with bounded contexts and data ownership declared, and the relevant ADRs.
• The data scope in question: a service's primary store, a shared store, a cross-context read model, or a cache tier.
• Primary access patterns: read shape, write shape, hot keys, and transactional groupings.

Optional: