GraphQL Schema & Resolver Conventions

graphqlapischemaresolvers

Conventions for designing GraphQL schemas and writing resolvers

Целевые файлы

Файл	Формат
.cursorrules	plaintext
CLAUDE.md	markdown

Содержимое

Use SDL-first schema design; keep all type definitions in .graphql files. Name queries with nouns (user, posts), mutations with verb+noun (createUser, deletePost). Always define input types for mutations; never use inline scalar arguments. Return payload objects from mutations: { success, errors, data } pattern. Implement cursor-based pagination (Relay spec) for all list fields. Never return null for list fields; return an empty array instead. Use DataLoader for all database calls inside resolvers to prevent N+1 queries. Add @deprecated(reason: "...") before removing any field; keep it one release cycle. Validate all inputs with custom scalars (Email, URL, UUID) where applicable. Keep resolvers thin: delegate business logic to service/domain layer functions. Version the schema via field deprecation, not URL versioning. Document every type and field with a description string in SDL. Limit query depth to ${max_depth} via server-side validation. Use persisted queries in production to prevent arbitrary query execution.

Переменные

ID	Метка	По умолчанию	Варианты
max_depth	Maximum query depth	10	—

CLI

npx mindaxis apply graphql-rules --target cursor --scope project

Используется в паках

GraphQL Development

Type-safe APIs with GraphQL

Открыть пак →

← Правила