What problem does it solve? Distributed systems built on SQS, EventBridge, and Lambda fail in predictable ways — duplicate deliveries, poison messages, retry storms, and dead dependencies — and this Skill provides concrete design guides so producers and consumers handle those failures instead of losing data silently. ## Core Features & Use Cases - Idempotency and Deduplication: Classify operations (SET/PUT vs INCREMENT/external side effects), choose business-level idempotency keys, and implement a conditional-write store with TTL. - Timeouts, Retries, and Circuit Breakers: Set explicit per-hop timeouts, classify retryable vs non-retryable errors, apply exponential backoff with jitter, and add circuit breakers and bulkheads for shared dependencies. - DLQ and Poison Message Handling: Configure DLQs with maxReceiveCount, alarm on queue depth, perform safe redrive after root-cause fixes, and use ReportBatchItemFailures for partial batch failures. - Use Case: When reviewing a Lambda consumer of an SQS queue, apply the checklists to verify the queue has a DLQ, the consumer is idempotent, and batch failures report only failed items. ## Quick Start Ask the assistant to review the resilience of an SQS consumer Lambda, covering idempotency keys, retry policy, and DLQ configuration.
