orm-reviewer_skill

Overview

This skill reviews ORM code for query efficiency, loading strategies, transaction safety, and model relationship patterns across Prisma, TypeORM, SQLAlchemy, Django ORM, and GORM. It highlights N+1 query sources, recommends eager vs lazy loading, checks transaction boundaries, and suggests model-level optimizations to reduce round-trips and data transfer. The goal is concrete, actionable fixes developers can apply rapidly.

How this skill works

The skill scans project files and idioms (e.g., schema.prisma, @Entity annotations, models.py, GORM struct tags) to detect ORM usage. It identifies patterns that cause N+1 queries, unnecessary full-row selects, missing transactions around multi-step operations, and suboptimal loading strategies, then produces prioritized findings with code-level fix suggestions. It focuses on runtime query behavior and usage patterns rather than raw SQL or overall schema design.

When to use it

• Performing an ORM-focused code review before a release
• Auditing repositories for N+1 queries and lazy-loading pitfalls
• Verifying transaction handling for multi-step workflows (orders, payments, inventory)
• Improving query efficiency and reducing payload sizes
• Assessing relation loading strategy choices (eager vs lazy) across services

Best practices

• Prefer explicit eager loading (include/preload/select_related) for list endpoints to prevent N+1
• Select only required columns and nested fields instead of fetching entire rows
• Wrap related multi-step operations in transactions to avoid partial state
• Use batch/bulk operations (createMany, bulk insert/update) where supported
• Choose selectinload/joinedload or ORM-specific prefetch strategies based on relation cardinality

Example use cases

• Scan a TypeORM service that loads users and prints orders to find lazy-loading N+1s and suggest relations or QueryBuilder joins
• Inspect Prisma resolvers to replace per-row findMany loops with include/select patterns and transaction wrappers
• Review SQLAlchemy code to swap lazy loads for selectinload/joinedload and recommend session scoping
• Audit Django views to replace repeated property access with select_related or prefetch_related and annotate counts
• Analyze GORM handlers to add Preload and Select clauses and identify opportunities for nested preloads