What problem does it solve? Standard Lambda functions lose all progress when interrupted, making long-running multi-step workflows fragile and hard to recover. This Skill guides you through building durable Lambda functions that checkpoint state, survive interruptions, and can run for up to one year with automatic retry and replay semantics. ## Core Features & Use Cases - Durable Execution Patterns: Implement steps, waits, callbacks, child contexts, and parallel/map operations using the durable-execution-sdk for TypeScript and Python. - Replay Model Guardrails: Enforce critical rules like keeping non-deterministic code inside steps, avoiding nested durable operations, and returning values instead of mutating closures. - Error Handling & Saga Pattern: Apply retry strategies, compensating transactions, circuit breakers, and unrecoverable error semantics for distributed workflows. - Testing & Deployment: Test locally with LocalDurableTestRunner or DurableFunctionTestRunner, and deploy with CloudFormation, CDK, or SAM using qualified ARNs. - Use Case: Build an order-processing workflow that reserves inventory, charges payment, waits up to 24 hours for human approval via callback, and automatically runs compensating refunds if any step fails. ## Quick Start Ask the AI to create a durable Lambda function in TypeScript that fetches user data in a step, waits five seconds, and then processes the result with automatic retries.