Use when reviewing Go code — covers goroutine leak detection, context propagation rules, error handling patterns (sentinel vs typed vs wrapped), slice/map pitfalls, interface design, nil traps, and common anti-patterns. Load alongside review-accuracy-calibration to ensure findings are well-calibrated before posting…
Use when reviewing or designing GraphQL and gRPC APIs — covers schema design and evolution, N+1/DataLoader batching, subscriptions, Protobuf field numbering, gRPC streaming modes, breaking-change detection, anti-patterns, and code review signals across TypeScript, Go, and Java.
Use when building or reviewing internationalized applications — covers message externalization, ICU MessageFormat, CLDR plural rules, Intl.DateTimeFormat/NumberFormat, RTL layout with CSS logical properties, Intl.Collator for locale-correct sorting, translation workflow, pseudo-localization testing, and i18n…
Use when reviewing Java code — covers null safety and Optional misuse, Stream API pitfalls, concurrency hazards (synchronized scope, CompletableFuture, ThreadLocal), resource management via try-with-resources, generics and type erasure traps, immutability patterns, and common Java anti-patterns. Load alongside…
Use when reviewing code for language-specific correctness — covers idiomatic patterns, common anti-idioms, and code review red flags for TypeScript/JavaScript, Python, Java, Go, Rust, and C++, with before/after examples and a quick reference table.
Use when designing or reviewing event-driven systems — covers Point-to-Point vs Pub/Sub, Competing Consumers, delivery guarantees (at-most-once, at-least-once, exactly-once), Transactional Outbox, Dead Letter Queues, Kafka partition design and consumer groups, message ordering and idempotency, backpressure, and…
Use when designing or reviewing microservice architectures — covers Saga (choreography vs orchestration), Bulkhead, API Gateway, Circuit Breaker, Retry with Backoff, Fallback and graceful degradation, Health Probes, Service Mesh/Sidecar, and resilience anti-patterns (retry storm, cascading failure, thundering herd)…
Use when planning or reviewing monolith decomposition, service extraction, or database migration — covers Strangler Fig, Anti-Corruption Layer, Branch by Abstraction, Database Decomposition (Expand-Contract, dual-write), Change Data Capture (Debezium, AWS DMS), and migration anti-patterns across TypeScript, Go, and…
Use when designing or reviewing multi-tenant systems — covers tenant isolation strategies (RLS, schema-per-tenant, DB-per-tenant), tenant context propagation, tenant-aware caching, tenant lifecycle management, shared vs isolated infrastructure tradeoffs, and multi-tenant anti-patterns across TypeScript, Go, Python…
Use when reviewing code for observability correctness — covers Structured Logging, Log Levels, Distributed Tracing, Metrics (counters/gauges/histograms), Health Checks, Alerting, Correlation IDs, and Audit Trails with red flags and fix strategies across TypeScript, Python, Java, and Go.
Use when you need end-to-end examples of detecting a code smell, identifying the underlying anti-pattern, selecting a refactoring technique, and applying the appropriate design pattern — each walkthrough traces the full diagnostic flow with before/after TypeScript code.
Use when reviewing a PR for performance regressions, when a service is slow under load, when database query times are spiking, or when CPU or memory usage is growing unexpectedly.
Use when starting a PR review and you need the end-to-end orchestration of all review skills into one complete workflow. This is the capstone skill that sequences every component skill into a principled, time-boxed process from PR load to verdict.
Python-specific code review guide for AI reviewers. Covers mutable defaults, type hints, GIL and concurrency patterns, import anti-patterns, Pythonic idioms, and exception handling. Load alongside review-accuracy-calibration to calibrate confidence before posting Python findings.
Use when designing or reviewing real-time communication — covers WebSocket lifecycle, Server-Sent Events, long polling, protocol selection, connection manager, horizontal scaling with Redis Pub/Sub, auth at upgrade, backpressure, and real-time anti-patterns with TypeScript and Go examples.
Use when methods are too long, contain duplicate code, or have complex expressions that need simplification — covers Extract Method, Inline Method, Extract Variable, and 6 more techniques.
Use when reviewing code for functional programming quality — covers Pure Functions, Immutability, Map/Filter/Reduce, Function Composition, Higher-Order Functions, Currying, Functors/Monads, and Pattern Matching with red flags and anti-patterns.
Use when managing inheritance hierarchies, extracting shared behavior, deciding between inheritance and delegation, or dealing with Refused Bequest, Parallel Inheritance, or Speculative Generality smells.
Use when responsibilities are in the wrong class, classes are too large or too small, or coupling between classes needs to change — covers Move Method, Extract Class, Hide Delegate, and 5 more techniques.
Use when data handling needs improvement — encapsulating fields, replacing primitives with objects, replacing type codes, managing associations, or fixing Data Class and Primitive Obsession smells.
Use when conditional logic is complex, nested, duplicated, or uses type codes instead of polymorphism — covers Decompose Conditional, Guard Clauses, Replace Conditional with Polymorphism, and 5 more techniques.
Use when method interfaces are confusing, parameters are excessive, or method responsibilities are unclear — covers Rename Method, Add/Remove Parameter, Introduce Parameter Object, Preserve Whole Object, and 9 more techniques.
Use when the user asks to refactor code, review code quality, or fix code smells — orchestrates the full refactoring pipeline from detection through fix, with parallel worktree isolation per independent refactoring group.
Use when you have identified a code smell and need to select the right refactoring technique, assess risk, and prioritize the work — maps all 23 code smells to specific fix paths, difficulty levels, and risk ratings.