Night Market Skill — ported from claude-night-market/archetypes. For the full experience with agents, hooks, and commands, install the Claude Code plugin.

The CQRS and Event Sourcing Paradigm

When To Use

• Designing event-sourced systems with complex domain logic
• Systems requiring full audit trails of state changes

When NOT To Use

• Simple CRUD applications without complex domain logic
• Small projects where event sourcing adds unnecessary complexity

When to Employ This Paradigm

• When read and write workloads have vastly different performance characteristics or scaling requirements.
• When all business events must be captured in a durable, immutable history or audit trail.
• When a business needs to rebuild projections of data or support temporal queries (e.g., "What did the state of this entity look like yesterday?").

Adoption Steps

1. Identify Aggregates: Following Domain-Driven Design principles, specify the bounded contexts and the business invariants that each command must enforce on an aggregate.
2. Model Commands and Events: Define the schemas and validation rules for all commands and the events they produce. Document a clear strategy for versioning and schema evolution.
3. Implement the Write Side (Command Side): Command handlers are responsible for loading an aggregate's event stream, executing business logic, and atomically appending new events to the stream.
4. Build Projections to the Read Side: Create separate read models (projections) that are fed by subscriptions to the event stream. Implement back-pressure and retry policies for these subscriptions.
5. validate Full Observability: Implement detailed logging that includes event IDs, sequence numbers, and metrics for tracking the lag time of each projection.

Key Deliverables

• An Architecture Decision Record (ADR) detailing the aggregates, the chosen event store technology, the projection strategy, and the expected data consistency model (e.g., eventual consistency SLAs).
• A suite of tests for command handlers that use in-memory event streams, complemented by integration tests for the projections.
• Operational tooling for replaying events, taking state snapshots for performance, and managing schema migrations.