| name | dynamodb-design |
| description | Design DynamoDB schemas from access patterns. Covers key design, GSIs, hot-partition mitigation, atomic write patterns (optimistic locking, counters, batch ops), DynamoDB Streams and CDC, schema evolution without downtime, and local testing. Stack-agnostic modeling with TypeScript code examples. |
DynamoDB Design
Opinionated patterns for designing a DynamoDB schema: inventory access patterns, derive keys, add GSIs, validate against scaling and cost constraints. Stack-agnostic — for CDK provisioning see aws-cdk-patterns.
When to load each reference
| Task | Reference file |
|---|
| Designing a new schema from scratch | references/00-methodology.md + references/01-modeling.md |
| Adding a new access pattern to an existing table | references/00-methodology.md (extension branch) + references/01-modeling.md |
| Migrating a schema without downtime | references/05-evolution.md |
| Throttling or uneven partition usage | references/02-scaling.md |
| Optimistic locking, atomic counters, or batch ops | references/03-write-correctness.md |
| Wiring DynamoDB Streams or CDC | references/04-streams-cdc.md |
| Running DynamoDB locally for tests | references/06-testing-local-dev.md |
| Diagnosing a production symptom | references/07-gotchas.md |
| Atomic uniqueness, identity-verified updates, cursor pagination (runtime patterns with full TypeScript) | aws-cdk-patterns/references/04-database.md §4-6 |
Conventions
- Code examples use
@aws-sdk/lib-dynamodb (AWS SDK for JavaScript v3). Patterns translate mechanically to boto3 (Python), the Go SDK, etc.
- Key schemas use
pk / sk as attribute names with sort-key prefix conventions documented per pattern.
- Every runtime code snippet is paired with a verification command when one applies.
Further reading
- Sibling skill:
aws-cdk-patterns — provisioning DynamoDB tables with CDK, Aurora patterns, and the three canonical runtime patterns referenced here.
