# AWS Architecture

> AWS architecture patterns for webhooks, event routing, SQS/Lambda pipelines, microservices, environment target guards, and long-running server deployment with CI/CD. Use when designing webhook ingestion, adding SQS queues, wiring Lambda consumers, deciding between FIFO vs standard queues, reviewing event-driven architecture, structuring Python Lambda microservice directories, guarding which environment of an external system a service calls, choosing between serverless and a long-running server, or deploying a containerized server to ECS Express Mode with push-to-deploy CI/CD.

- Skill: `jd-solanki/aws-architecture` (Agent Skill, multi-file: 11 files)
- Install (CLI): `npx skillmds@latest add jd-solanki/aws-architecture`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jd-solanki/aws-architecture/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: jd-solanki (https://skillmd.com/u/jd-solanki)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jd-solanki/aws-architecture

---


# AWS Architecture

Architecture patterns for AWS — serverless and long-running.

- [Webhook Ingestion Architecture](webhooks/WEBHOOK_ARCHITECTURE.md) — Webhook pipeline: API GW → SQS FIFO → Routing Lambda → per-type SQS → Handler Lambda. Includes queue naming, payload routing, DLQ strategy, and reference diagrams.
- [Microservice Architecture](microservices/MICROSERVICE_ARCHITECTURE.md) — Recommended directory layout and layer sharing strategy for Python Lambda microservices.
- [Environment Target Guard](environment-guard/ENVIRONMENT_GUARD_ARCHITECTURE.md) — Alarm when a service calls the wrong environment of an external system, staging instead of production or the reverse: the shared client logs its target, a CloudWatch metric filter counts calls to the wrong one, a standing alarm and a clock-driven reminder alarm notify SNS. Adds a CloudFormation hook as a second layer that catches a flip deployed to an idle function, which no runtime check can see. Ships copy-paste nested stacks for both, composes with queue-triggered-lambda.yml, and includes how to rehearse the alarms from the CLI and with a canary function.
- [Long-Running Server Deployment](deployment/DEPLOYMENT_ARCHITECTURE.md) — Deploy pipeline: push → test → image to ECR → migrate → ECS Express Mode (Fargate + ALB), with durable infrastructure in a stack the pipeline never runs. Includes why not serverless or EC2, disposable compute, keyless OIDC, reversible-before-irreversible ordering, and expand/contract migrations.