event-driven-architect_skill

Overview

This skill designs event-driven architectures focused on scalable, loosely-coupled microservices using Kafka, RabbitMQ, event sourcing, CQRS, and saga patterns. It helps teams choose patterns, define event schemas, and map flows for reliable asynchronous communication. The guidance balances operational concerns like throughput, retention, and monitoring with domain modeling for events and aggregates.

How this skill works

The skill analyzes your domain and transaction boundaries, then recommends patterns (event sourcing, CQRS, sagas) and messaging infrastructure (topics, partitions, retention, schema registry). It provides concrete artifacts: event contracts (Avro/Protobuf), sample Kafka/RabbitMQ configuration, read/write model separation, and orchestration strategies for sagas. It also highlights operational controls—idempotency, dead-letter queues, monitoring, and event versioning—to ensure resilience and observability.

When to use it

• Designing scalable microservices that must be loosely coupled
• Implementing event sourcing to store state as immutable events
• Setting up Kafka, RabbitMQ, or real-time event streams
• Applying CQRS to separate read and write responsibilities
• Coordinating distributed transactions via the saga pattern
• Building real-time data pipelines or publish/subscribe systems

Best practices

• Version all event schemas and maintain a schema registry (Avro/Protobuf)
• Make handlers idempotent and record processing offsets to avoid duplicate effects
• Design read models as optimized projections for queries, separate from the write model
• Use dead-letter queues and retry policies for transient failures
• Instrument correlation IDs, monitor event lag, and track throughput
• Model sagas with explicit compensation steps and persistent saga state

Example use cases

• E-commerce order flow: use event sourcing for order state, projections for queries, and a saga to reserve inventory and process payment
• Real-time analytics: stream events from game engines into Kafka topics, project into OLAP stores for dashboards
• Distributed billing: commands create invoice events, read models power customer billing views, and sagas ensure payment consistency
• Inventory system: publish stock change events to multiple microservices with idempotent handlers and DLQs
• Multiplayer game state sync: event streams distribute player actions and projections build leaderboards