# AWS Best Practices

> AWS service-specific best practices and configuration rules. Use when making decisions about RDS environment isolation, SSM Parameter Store, SQS consumer types, Lambda function structure, or SAM template conventions.

- Skill: `jd-solanki/aws-best-practices` (Agent Skill, multi-file: 12 files)
- Install (CLI): `npx skillmds@latest add jd-solanki/aws-best-practices`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jd-solanki/aws-best-practices/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-best-practices

---


# AWS Best Practices

Service-specific configuration and lifecycle rules. Each reference covers one AWS service in depth.

- [RDS.md](RDS.md) — RDS/Aurora: one host per environment, why prod must not share an instance with staging, staging fidelity (engine flavour and version), and sizing staging down instead of sharing
- [SSM.md](SSM.md) — Parameter Store: when to create parameters, runtime vs deploy-time fetching, the bootstrap pattern, and the decision tree for secrets vs config
- [SQS.md](SQS.md) — SQS consumer types, using shared layer TypedDict types in Lambda handlers
- [LAMBDA.md](LAMBDA.md) — Lambda handler structure: SQS consumption pattern, `batchItemFailures`, `process_record()` separation, `setup_logger()` / `log_prefix_scope()`, event logging, CloudWatch alarms, and Lambda/SQS naming
- [SAM.md](SAM.md) — SAM template conventions: no explicit physical resource names (auto-generate to keep deploys safe), resource grouping by topic, `Fn::ToJsonString`, `LAMBDA_ENV_VARS` pattern
- [templates/](templates/) — copy-paste nested-stack templates referenced above: `lambda-cloudwatch-alarm.yml`, `dlq.yml`, `queue-triggered-lambda.yml`
- [shared_layer/](shared_layer/) — copy-paste shared-layer modules referenced above: `log.py`, `utils/string.py`, `utils/encoders.py`