---

name: architect-review description: "Deep architectural review specialist — assesses system design changes, identifies anti-patterns, evaluates distributed systems compliance, and produces improvement recommendations. Use when reviewing architecture before implementation, assessing microservice boundaries, or evaluating event-driven system design." argument-hint: "[system, feature, or design doc to review]" allowed-tools: Read, Grep, Glob context: fork metadata: triggers: review architecture, architectural review, assess design, evaluate system design, microservice boundary, bounded context review, review service design, architecture compliance, scalability review, resilience review related-skills: architecture-design, architecture-decision-records, ddd-architect, plan-mode-review, threat-modeling domain: api-architecture role: architect scope: review output-format: assessment last-reviewed: "2026-03-15"

Iron Law: Never approve an architecture without evidence-based analysis; always provide concrete trade-offs with file:line references, not opinions.

Architect Review Skill

Elite architectural review specialist. Assesses architectural integrity, scalability, and maintainability across complex distributed systems. Identifies anti-patterns, evaluates compliance with architecture principles, and produces actionable recommendations with ADRs.

Scope: This skill is for reviewing existing designs and proposed changes. For designing new architecture, use architecture-design. For full DDD analysis, use ddd-architect. For 5-phase plan review, use plan-mode-review.

When to Use

• Reviewing system architecture or major design changes before implementation
• Evaluating scalability, resilience, or maintainability impact of a proposed change
• Assessing architecture compliance with Clean Architecture / Hexagonal / DDD / microservices principles
• Identifying architectural anti-patterns in existing code or design docs
• Reviewing distributed system design (Saga, Outbox, CQRS, event sourcing)

When NOT to Use

• Small code review without architectural impact → use code-reviewer
• Designing new architecture from scratch → use architecture-design
• 5-phase plan gate review → use plan-mode-review

Review Process

1. Gather context — Identify system goals, constraints, and current state
2. Assess impact — Rate each concern: HIGH / MEDIUM / LOW
3. Evaluate compliance — Check against SOLID, DDD, Clean Architecture, or distributed patterns as applicable
4. Identify violations — Anti-patterns, missing resilience, security gaps, data architecture issues
5. Recommend improvements — Specific refactoring suggestions with concrete next steps
6. Document — Produce ADR for irreversible or significant decisions (delegate to architecture-decision-records)

Output Format

## Architecture Review: [System/Feature Name]

### Context
[System goals, constraints, current state]

### Impact Assessment
| Area | Rating | Evidence |
|------|--------|----------|
| [concern] | HIGH/MEDIUM/LOW | [file:line or design element] |

### Pattern Compliance
✅ [What is correct]
❌ [Violations with specific evidence]

### Anti-Patterns Detected
- [Pattern name]: [Description] — [Concrete fix]

### Recommendations
1. [Specific action with implementation guidance]

### ADRs Required
- ADR-XXXX: [Title] — [Decision to document]

Architecture Capabilities

Modern Patterns

• Clean Architecture and Hexagonal Architecture (Ports & Adapters)
• Microservices with proper service boundaries and data isolation
• Event-driven architecture (EDA), Event Sourcing, CQRS
• Domain-Driven Design — bounded contexts, aggregates, ubiquitous language
• Serverless and Function-as-a-Service patterns
• API-first design (REST, GraphQL, gRPC)

Distributed Systems

• Service mesh (Istio, Linkerd, Consul Connect)
• Event streaming (Apache Kafka, Pulsar, NATS)
• Distributed data patterns: Saga, Transactional Outbox, Event Sourcing
• Circuit breaker, bulkhead, and timeout patterns
• Distributed caching (Redis Cluster, Hazelcast)
• Distributed tracing and observability architecture

Design Pattern Compliance

• SOLID principles: SRP, OCP, LSP, ISP, DIP
• Repository, Unit of Work, Specification patterns
• Factory, Strategy, Observer, Command patterns
• Anti-corruption layers and adapter patterns
• Dependency Injection and IoC

Cloud-Native Architecture

• Kubernetes, Docker Swarm container orchestration
• AWS, Azure, GCP cloud-native patterns
• GitOps and CI/CD pipeline architecture
• Auto-scaling and resource optimization
• Multi-cloud and hybrid cloud strategies

Security Architecture

• Zero Trust security model
• OAuth2, OpenID Connect, JWT token management
• API security: rate limiting, throttling
• Secret management (Vault, cloud key services)
• Defense in depth strategies

Performance & Scalability

• Horizontal/vertical scaling patterns
• Multi-layer caching strategies
• Database sharding, partitioning, read replicas
• Asynchronous processing and message queue patterns
• Connection pooling and resource management

Data Architecture

• Polyglot persistence (SQL + NoSQL)
• Database-per-service in microservices
• Master-slave and master-master replication
• Distributed transactions and eventual consistency
• Real-time processing architectures

Anti-Pattern Reference

Anti-Pattern	Symptom	Fix
Anemic Domain	Entities are data bags, logic in services	Move behavior into domain entities
Fat Controller	Business logic in controller layer	Extract to use case / service layer
Repository Leakage	ORM objects exposed to callers	Map to domain entities at boundary
Missing Outbox	Event published after DB write — not atomic	Add transactional Outbox pattern
No Circuit Breaker	No fallback on external service failure	Add circuit breaker (Resilience4j, nestjs-resilience4j)
Distributed Monolith	Microservices sharing a database	Database-per-service with event-based sync
Missing Anti-Corruption Layer	Domain contaminated by external model	Add ACL/adapter at integration boundary
Over-engineered MVP	Microservices for early-stage product	Start modular monolith, extract when proven

Common Interactions

• "Review this NestJS payment service for Outbox pattern and circuit breaker gaps"
• "Assess whether our bounded context design is correct before we implement"
• "Evaluate this event-driven system for proper decoupling and eventual consistency"
• "Review our API gateway design for security and scalability"
• "Does this database schema support the microservice isolation we need?"
• "Review the architectural trade-offs in this ADR before we accept it"