aws-well-architected-framework-review — команда для Claude Code: обзор
Команда aws-well-architected-framework-review для Claude Code проверяет AWS-нагрузку по столпам Well-Architected Framework и связывает риски с кодом и инфраструктурой. Разбирает IaC, конфигурации и архитектуру приложения.
MIT-0 ★ 280 обновлён 2026-10-06
.claude/commands/aws-well-architected-framework-review.mdЧто делает aws-well-architected-framework-review
Сначала команда определяет границы нагрузки, затем изучает IaC, настройки развёртывания и код приложения. Она проверяет безопасность, надёжность, производительность, стоимость, устойчивость и операционное управление. Для специализированной нагрузки, например serverless или IoT, учитываются вопросы соответствующей линзы Well-Architected.
Результат включает схему архитектуры в PlantUML и оценки с доказательствами: путями файлов и номерами строк, пробелами и связанными рисками. Находки ранжируются по влиянию и вероятности. Если проверяется описание архитектуры без кода, неподтверждённые детали помечаются как требующие проверки в коде.
Команда требует Claude Code. Установка выполняется копированием файла в .claude/commands/aws-well-architected-framework-review.md; для анализа IaC и кода нужны соответствующие файлы проекта, но обзор может опираться и на описание архитектуры.
Какие задачи решает
- проверить IaC и настройки развёртывания AWS
- найти пробелы в IAM, шифровании и сетевой защите
- сопоставить архитектурные риски с файлами и строками кода
- составить схему компонентов и потоков данных в PlantUML
- расставить приоритеты исправлений по влиянию и вероятности
Примеры запросов агенту
- Проведи обзор AWS-нагрузки по Well-Architected Framework для этого проекта.
- Проверь IaC и конфигурации развёртывания, найди риски безопасности и надёжности.
- Составь PlantUML-схему архитектуры и приоритизируй найденные проблемы.
Как установить aws-well-architected-framework-review
Вручную: положить файлы
.claude/commands/aws-well-architected-framework-review.mdВызов — /aws-well-architected-framework-review. Для всех проектов — ~/.claude/commands/
Вручную: положить файлы
~/.codex/prompts/aws-well-architected-framework-review.mdВызов — /prompts:aws-well-architected-framework-review. В Codex этот формат помечен устаревшим, замена — скиллы
Вручную: положить файлы
.github/prompts/aws-well-architected-framework-review.prompt.mdВручную: положить файлы
.gemini/commands/aws-well-architected-framework-review.tomlGemini хранит команды в TOML: текст запроса — в поле prompt
Вопросы об aws-well-architected-framework-review
Как установить aws-well-architected-framework-review в Claude Code?
Скопируйте файл команды в .claude/commands/aws-well-architected-framework-review.md.
Что делает aws-well-architected-framework-review?
Команда анализирует код, IaC и конфигурации AWS-нагрузки, создаёт PlantUML-схему и формирует подтверждённые находки с приоритетами исправления.
Можно ли провести обзор без исходного кода?
Да. Команда может опираться на описание архитектуры; детали реализации, которые нельзя проверить, отмечаются как требующие проверки в коде.
Исходный текст
aws-well-architected-framework-review.md · MIT-0
Perform a full AWS Well-Architected Framework review of the user's workload by analyzing code, IaC, and configurations to produce evidence-backed findings with prioritized remediation.
## Step 1: Define the workload scope
Ask the user to describe the workload:
> What workload would you like me to review? Please share:
> - **Workload name** and brief description
> - **Code packages/directories** to analyze (IaC, application code, CI/CD configs)
> - **Business criticality** (critical, high, standard, low)
> - **Current pain points** (optional — anything you already know is problematic)
If the user has already provided architecture details or you are in a codebase with IaC, skip the prompt and proceed with discovery.
**IMPORTANT**: When no code or IaC is available to analyze (e.g., the user describes their architecture verbally), proceed with the review based on the information provided. Produce the full report using the architecture description as evidence. Mark findings where you cannot verify implementation details as "Based on description — verify in code." Do NOT ask for code if the user has already given you enough context to perform a meaningful review.
Determine if a specialized WA Lens applies:
- SaaS, Serverless, Data Analytics, Machine Learning, IoT, Containers, Games, Financial Services, Healthcare
If a lens is obvious from the code (e.g., Lambda-heavy → serverless), note it and apply lens-specific questions.
## Step 2: Infrastructure Discovery
Analyze all infrastructure-as-code and deployment configurations in the codebase.
You MUST examine:
- CDK code (TypeScript, Python, Java, Go)
- CloudFormation templates (YAML, JSON)
- Terraform configurations (.tf files)
- SAM/Serverless Framework templates
- CI/CD pipeline definitions (CodePipeline, GitHub Actions, etc.)
- Monitoring configurations (CloudWatch alarms, dashboards)
- Deployment configurations (CodeDeploy, ECS deployment settings)
For each infrastructure component, document:
- Resource type, logical name, and configuration
- File path and line numbers where defined
- Security-relevant configs (IAM, encryption, network)
- Resilience configs (multi-AZ, backups, scaling)
- Cost-relevant configs (instance types, capacity mode)
You MUST create an architecture diagram in PlantUML showing:
- All major components and their relationships
- Data flows and external dependencies
- Trust and network boundaries
## Step 3: Application Architecture Discovery
Analyze application code for architectural patterns:
- Entry points (API handlers, event processors, scheduled tasks)
- Service communication patterns (sync/async, retries, timeouts, circuit breakers)
- Data access patterns (queries, caching, connection management)
- Error handling and resilience patterns
- Authentication/authorization logic
- Observability instrumentation (logging, tracing, metrics)
## Step 4: Evaluate each pillar with code evidence
For each pillar, assess against the WA Framework questions. Every assessment MUST include:
- **Status**: "Implemented", "Partially Implemented", "Not Implemented", "Cannot Determine"
- **Evidence**: specific file paths and line numbers
- **Gaps**: what's missing or could be improved
- **Risk**: what could go wrong due to the gap
### Operational Excellence
- **OPS 4 — Observability**: Look for structured logging, distributed tracing, custom metrics. Evidence: log statements, X-Ray/OTEL instrumentation.
- **OPS 5 — Reduce defects**: Look for automated testing, linting, staged deployments. Evidence: test files, CI/CD configs.
- **OPS 6 — Deployment risk**: Look for canary/blue-green, feature flags, rollback automation. Evidence: deployment config, CodeDeploy settings.
- **OPS 8 — Workload health**: Look for health dashboards, composite alarms. Evidence: alarm definitions, health checks.
- **OPS 10 — Event management**: Look for event routing, automated remediation. Evidence: EventBridge rules, Lambda remediations.
### Security
- **SEC 3 — Permissions**: Look for least privilege, permission boundaries. Flag any `"Action": "*"` or `"Resource": "*"` on mutating actions.
- **SEC 5 — Network protection**: Look for VPC design, security groups, WAF, VPC endpoints. Flag `0.0.0.0/0` on non-443/80 ports.
- **SEC 8 — Data at rest**: Look for encryption on ALL stores. Flag any storage resource without encryption.
- **SEC 9 — Data in transit**: Look for TLS enforcement. Flag TLS < 1.2.
- **SEC 2/3 — Identity**: Look for IAM roles, trust relationships, credential management. Flag hardcoded secrets.
### Reliability
- **REL 2 — Network topology**: Look for multi-AZ deployment, subnet distribution.
- **REL 3 — Demand adaptation**: Look for auto-scaling, capacity modes. Flag compute without scaling.
- **REL 4 — Distributed system interactions**: Look for retries with backoff, timeouts, idempotency.
- **REL 9 — Data backup**: Look for automated backups, PITR. Flag stateful resources without backups.
- **REL 11 — Component failures**: Look for redundancy, failover, health-based routing.
- **REL 13 — Disaster recovery**: Look for cross-region resources, DR automation.
### Performance Efficiency
- **PERF 1 — Resource selection**: Look for instance type justification, Graviton usage.
- **PERF 2 — Compute solution**: Look for right-sizing, memory/CPU alignment.
- **PERF 4 — Database solution**: Look for read replicas, connection pooling (RDS Proxy, DAX).
- **PERF 5 — Networking**: Look for CloudFront, VPC endpoints, edge caching.
- **PERF 8 — Tradeoffs**: Look for caching, async patterns, eventual consistency.
### Cost Optimization
- **COST 4 — Decommission resources**: Look for lifecycle policies, retention configs. Flag "never expire" logs.
- **COST 5 — Service selection**: Look for pricing model alignment. Flag over-provisioned resources.
- **COST 6 — Right-sizing**: Look for scaling boundaries, capacity allocation.
- **COST 8 — Data transfer**: Look for VPC endpoints vs NAT Gateway, cross-region patterns.
### Sustainability
- **SUS 2 — User behavior patterns**: Look for scale-to-zero, scheduled scaling.
- **SUS 4 — Data patterns**: Look for tiering, lifecycle management, compression.
- **SUS 5 — Hardware patterns**: Flag x86 where Graviton is available. Note managed service usage.
## Step 5: Risk Assessment
For each finding, assess using Impact × Likelihood:
**Impact**: Minor (limited blast radius) | Moderate (subset of users affected) | Severe (full outage, data loss, regulatory violation)
**Likelihood**: Low (specific conditions required) | Medium (possible under normal operations) | High (common failure mode, weak controls)
| Impact | Likelihood | Risk Level |
|----------|------------|------------|
| Severe | High | Critical |
| Severe | Medium | High |
| Severe | Low | High |
| Moderate | High | High |
| Moderate | Medium | Medium |
| Moderate | Low | Medium |
| Minor | High | Medium |
| Minor | Medium | Low |
| Minor | Low | Low |
Identify cross-pillar conflicts:
- Security controls that impact performance
- Cost optimizations that reduce reliability
- Reliability patterns that increase cost
## Step 6: Produce the report
Output a structured report:
```markdown
# Well-Architected Review: {Workload Name}
## Executive Summary
- **Date**: {date}
- **Workload**: {name}
- **Business Criticality**: {level}
- **Lens Applied**: {lens or "General"}
- **Packages Analyzed**: {list}
- **Findings**: {X} Critical, {Y} High, {Z} Medium, {W} Low
- **Overall Maturity**: {1-5} — {one-line justification}
## Architecture Overview
{PlantUML diagram}
{Brief description of architecture, key services, data flows}
## Pillar Scorecard
| Pillar | Score (1-5) | Key Strength | Key Gap |
|--------|-------------|--------------|---------|
| Operational Excellence | {score} | {strength} | {gap} |
| Security | {score} | {strength} | {gap} |
| Reliability | {score} | {strength} | {gap} |
| Performance Efficiency | {score} | {strength} | {gap} |
| Cost Optimization | {score} | {strength} | {gap} |
| Sustainability | {score} | {strength} | {gap} |
## Critical and High Risk Findings
{For each: ID, pillar, title, description, evidence (file:line), impact assessment, recommendation, **Fix:** block — a minimal ready-to-copy snippet in the workload's IaC dialect (AWS CLI when no IaC), using the actual resource names from the evidence; shown in the report only, never applied — effort, AWS services}
## Medium Risk Findings
{Same format, condensed — each finding still carries its **Fix:** block}
## Low Risk Findings
{Summary table: ID | Pillar | Title | Recommendation}
## Cross-Pillar Trade-offs
{Conflicts between pillars and recommended resolution}
## Prioritized Remediation Plan
### Quick Wins (< 1 week)
| Finding | Action | Impact | Effort |
|---------|--------|--------|--------|
{Config changes, enabling features, adding tags/alarms}
### Foundation (1-4 weeks)
| Finding | Action | Impact | Effort | Dependencies |
|---------|--------|--------|--------|--------------|
{Multi-AZ, CI/CD improvements, monitoring, caching}
### Strategic (1-3 months)
| Finding | Action | Impact | Effort | Dependencies |
|---------|--------|--------|--------|--------------|
{DR, re-architecture, compliance programs, advanced automation}
## Next Steps
{Top 5 concrete actions the team should take this week}
```
## Step 6b: Emit structured output (`aws-well-architected-framework-review.json`)
After the markdown report, ALSO emit a machine-readable `aws-well-architected-framework-review.json` so the review can be
diffed over time (a CI gate), aggregated across workloads, or imported into the WA Tool. Conform
to the versioned contract in `schemas/aws-well-architected-framework-review-v1.schema.json` (repo root). Serialize the same
findings you just reported — do not re-run the analysis.
**Rules:**
- One `findings` entry per BP you evaluated, including `implemented` and `not_applicable` ones, so
a consumer can tell an evaluated-and-passed BP from one never assessed.
- For a core-framework or ID-bearing lens BP, `bp_id` MUST be canonical `PILLAR##-BP##` (e.g.
`SEC03-BP02`, `IOTCOST01-BP01`) — it is the identity key a baseline diff pairs on.
- Some lenses are topic-organized and expose no BP ID (serverless-applications, saas, government,
healthcare-industry, container-build, sap, streaming-media, plus a few mixed cases). For those,
OMIT `bp_id` and give a `title` (plus `lens`). They are advisory: reported, never gated. Do not
invent an ID to satisfy the diff.
- Set `review_mode` to the depth you ran (`full`, `quick`, `pillar-scoped`, `score`) and set
`skill_version` plus a unique `run_id`.
- Every gap (`not_implemented` / `partially_implemented`) MUST carry a `severity`; the gate ranks
by it, so an unrated gap fails the build closed. Omit severity only for non-gap statuses.
- `recall_note` MUST state coverage is high-recall but not exhaustive, so a gate never reads a
missing finding as proof a control exists.
- `recommendation` stays prose guidance, never a code diff.
```json
{
"schema_version": "1.0.0",
"workload": "{name}",
"date": "{YYYY-MM-DD}",
"review_mode": "full",
"skill_version": "{skill version}",
"run_id": "{date}T{time}-{workload-slug}",
"pillar_scores": { "security": 2, "reliability": 2 },
"findings": [
{
"bp_id": "SEC08-BP01",
"pillar": "security",
"status": "not_implemented",
"severity": "high",
"evidence": { "file": "infra/s3.tf", "line": 14 },
"effort": "low",
"recommendation": "Enable SSE and BlockPublicAccess on the uploads bucket."
}
],
"recall_note": "Full review. High recall but not exhaustive; absence of a finding is not proof of implementation."
}
```
Commit this file as `.well-architected/baseline.json` and run `tools/wa-ci` in a pipeline to gate
future PRs on the Well-Architected delta.
## Step 7: Offer follow-up
After delivering the report, offer:
> Would you like me to:
> - Deep-dive into a specific pillar with expanded analysis?
> - Generate IaC templates to remediate a specific finding?
> - Create a migration plan for a specific architectural change?
> - Compare your workload against a specific WA Lens in detail?
> - Generate automated checks (Config rules, custom metrics) for ongoing compliance?
> - Set up a continuous WA gate: commit `aws-well-architected-framework-review.json` as a baseline and run `tools/wa-ci` in CI?
> - Produce a WA Tool import for tracking in the AWS console?
## Calibration Guidance
- A workload with multi-AZ, encryption, CI/CD with rollback, monitoring, and auto-scaling is MATURE — most findings should be improvements, not Critical
- Do NOT manufacture Critical findings for a well-built workload — accuracy over alarm
- When business criticality is "low"/"standard", accept simpler architectures (single-region is fine for internal tools)
- When business criticality is "critical", apply stricter standards (multi-region DR, chaos testing, sub-minute RTO expected)
- Every finding MUST have code evidence — no generic recommendations without backing
- If something cannot be determined from code, say "Cannot Determine" and explain what runtime/interview data is needed
- Acknowledge strengths prominently — a mature workload should feel validated, not just criticized
<!--
Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
SPDX-License-Identifier: MIT-0
-->
© aws-samples · MIT-0 · aws-samples/sample-well-architected-skills-and-steering