N+1 Query Detection Skill

Scan repository and service code for N+1 query patterns that cause performance degradation.

Trigger Conditions

• Repository or service files are modified
• Performance review is requested
• New data access patterns are introduced
• User invokes with "N+1 check" or "n-plus-1-query-detection"

Input Contract

• Required: Path to repository and/or service files
• Optional: GORM configuration for preload analysis

Output Contract

• List of N+1 patterns found with file:line references
• Severity assessment (critical for high-volume paths, low for admin-only)
• Fix recommendation (Preload, Joins, or batch query)
• Unbounded query warnings (missing Limit)

Tool Permissions

• Read: Repository files, service files, model relationship definitions
• Write: None (read-only analysis)
• Search: Grep for loop patterns,

  .Find()

  ,

  .First()

  , relationship access

Execution Steps

1. Scan for loops with queries: Find patterns where a database query is inside a

   for

   loop
2. Check preloads: For each query that accesses relationships, verify

   Preload()

   or

   Joins()

   is used
3. Check unbounded queries: Find

   .Find()

   calls without

   .Limit()

   on list endpoints
4. Assess severity: Rate each finding by the expected data volume (high-traffic endpoints are critical)
5. Recommend fixes: Suggest specific GORM patterns (

   Preload

   ,

   Joins

   ,

   SelectinLoad

   )
6. Report: Produce findings with code examples for the fix

Common Patterns (Anti-patterns)

// N+1: Query inside loop
users, _ := db.Find(&users)
for _, u := range users {
    db.Where("user_id = ?", u.ID).Find(&orders) // N+1!
}

// Fix: Preload
db.Preload("Orders").Find(&users)

References

• .cursor/rules/122-gorm-conventions.mdc

• .cursor/rules/047-performance-optimization.mdc