vtex/skills

vtex-io-graphql-api

Apply when working with GraphQL schema files in graphql/ or implementing resolvers in node/resolvers/ for VTEX IO apps.

View source
Original skill document

Rendered from the source repository. Headings, examples, code, tables, links, and referenced images are preserved.

GraphQL Schemas & Resolvers

When this skill applies

Use this skill when your VTEX IO app needs to expose a GraphQL API — either for frontend React components to query, for other VTEX IO apps to consume, or for implementing custom data aggregation layers over VTEX Commerce APIs.

  • Defining schemas in .graphql files in the /graphql directory
  • Writing resolver functions in TypeScript in /node/resolvers/
  • Configuring @cacheControl and @auth directives
  • Wiring resolvers into the Service class

Do not use this skill for:

  • Backend service structure and client system (use vtex-io-service-apps instead)
  • Manifest and builder configuration (use vtex-io-app-structure instead)
  • MasterData integration details (use vtex-io-masterdata instead)

Decision rules

  • The graphql builder processes .graphql files in /graphql and merges them into a single schema.
  • Split definitions across multiple files for maintainability: schema.graphql for root types, directives.graphql for directive declarations, types/*.graphql for custom types.
  • Use @cacheControl(scope: PUBLIC, maxAge: SHORT|MEDIUM|LONG) on all public Query fields. PUBLIC = shared CDN cache, PRIVATE = per-user cache.
  • Use @auth on all Mutations and on Queries that return sensitive or user-specific data.
  • Never use @cacheControl on Mutations.
  • Resolver function keys in the Service entry point MUST exactly match the field names in schema.graphql.
  • Always use ctx.clients in resolvers for data access — never raw HTTP calls.

Recommended directory structure:

text
graphql/
├── schema.graphql        # Query and Mutation root type definitions
├── directives.graphql    # Custom directive declarations (@cacheControl, @auth)
└── types/
    ├── Review.graphql    # Custom type definitions
    └── Product.graphql   # One file per type for organization

Built-in directives:

  • `@cacheControl`: scope (PUBLIC/PRIVATE), maxAge (SHORT 30s, MEDIUM 5min, LONG 1h)
  • `@auth`: Enforces valid VTEX authentication token. Without it, unauthenticated users can call the endpoint.
  • `@smartcache`: Automatically caches query results in VTEX infrastructure.

Hard constraints

Constraint: Declare the graphql Builder

Any app using .graphql schema files MUST declare the graphql builder in manifest.json. The graphql builder interprets the schema and registers it with the VTEX IO runtime.

Why this matters

Without the graphql builder declaration, the /graphql directory is completely ignored. Schema files will not be processed, resolvers will not be registered, and GraphQL queries will return "schema not found" errors. The app will link without errors but GraphQL will silently not work.

Detection

If you see .graphql files in a /graphql directory but the manifest does not include "graphql": "1.x" in builders, STOP and add the builder declaration.

Correct

json
{
  "builders": {
    "node": "7.x",
    "graphql": "1.x"
  }
}

Wrong

json
{
  "builders": {
    "node": "7.x"
  }
}

Missing "graphql": "1.x" — the /graphql directory with schema files is ignored. GraphQL queries return errors because no schema is registered. The app links successfully, masking the problem.


Constraint: Use @cacheControl on Public Queries

All public-facing Query fields (those fetching data that is not user-specific) MUST include the @cacheControl directive with an appropriate scope and maxAge. Mutations MUST NOT use @cacheControl.

Why this matters

Without @cacheControl, every query hits your resolver on every request — no CDN caching, no edge caching, no shared caching. This leads to unnecessary load on VTEX infrastructure, slow response times, and potential rate limiting. For public product data, caching is critical for performance.

Detection

If a Query field returns public data (not user-specific) and does not have @cacheControl, warn the developer to add it. If a Mutation has @cacheControl, STOP and remove it.

Correct

graphql
type Query {
  reviews(productId: String!, limit: Int): [Review]
    @cacheControl(scope: PUBLIC, maxAge: SHORT)

  productMetadata(slug: String!): ProductMetadata
    @cacheControl(scope: PUBLIC, maxAge: MEDIUM)

  myReviews: [Review]
    @cacheControl(scope: PRIVATE, maxAge: SHORT)
    @auth
}

type Mutation {
  createReview(review: ReviewInput!): Review @auth
}

Wrong

graphql
type Query {
  reviews(productId: String!, limit: Int): [Review]

  myReviews: [Review]
}

type Mutation {
  createReview(review: ReviewInput!): Review
    @cacheControl(scope: PUBLIC, maxAge: LONG)
}

No cache control on queries (every request hits the resolver), missing @auth on user-specific data, and @cacheControl on a mutation (makes no sense).


Constraint: Resolver Names Must Match Schema Fields

Resolver function keys in the Service entry point MUST exactly match the field names defined in schema.graphql. The resolver object structure must mirror the GraphQL type hierarchy.

Why this matters

The GraphQL runtime maps incoming queries to resolver functions by name. If the resolver key does not match the schema field name, the field will resolve to null without any error — a silent failure that is extremely difficult to debug.

Detection

If a schema field has no matching resolver key (or vice versa), STOP. Cross-check every Query and Mutation field against the resolver registration in node/index.ts.

Correct

graphql
type Query {
  reviews(productId: String!): [Review]
  reviewById(id: ID!): Review
}
typescript
// node/index.ts — resolver keys match schema field names exactly
export default new Service({
  graphql: {
    resolvers: {
      Query: {
        reviews: reviewsResolver,
        reviewById: reviewByIdResolver,
      },
    },
  },
})

Wrong

typescript
// node/index.ts — resolver key "getReviews" does not match schema field "reviews"
export default new Service({
  graphql: {
    resolvers: {
      Query: {
        getReviews: reviewsResolver,    // Wrong! Schema says "reviews", not "getReviews"
        getReviewById: reviewByIdResolver, // Wrong! Schema says "reviewById"
      },
    },
  },
})

Both fields will silently resolve to null. No error in logs.

Preferred pattern

Add the GraphQL builder to manifest:

json
{
  "builders": {
    "node": "7.x",
    "graphql": "1.x"
  }
}

Define the schema:

graphql
type Query {
  reviews(productId: String!, limit: Int, offset: Int): ReviewsResponse
    @cacheControl(scope: PUBLIC, maxAge: SHORT)

  review(id: ID!): Review
    @cacheControl(scope: PUBLIC, maxAge: SHORT)
}

type Mutation {
  createReview(input: ReviewInput!): Review @auth
  updateReview(id: ID!, input: ReviewInput!): Review @auth
  deleteReview(id: ID!): Boolean @auth
}

Define custom types:

graphql
type Review {
  id: ID!
  productId: String!
  author: String!
  rating: Int!
  title: String!
  text: String!
  createdAt: String!
  approved: Boolean!
}

type ReviewsResponse {
  data: [Review!]!
  total: Int!
  hasMore: Boolean!
}

input ReviewInput {
  productId: String!
  rating: Int!
  title: String!
  text: String!
}

Declare directives:

graphql
directive @cacheControl(
  scope: CacheControlScope
  maxAge: CacheControlMaxAge
) on FIELD_DEFINITION

enum CacheControlScope {
  PUBLIC
  PRIVATE
}

enum CacheControlMaxAge {
  SHORT
  MEDIUM
  LONG
}

directive @auth on FIELD_DEFINITION
directive @smartcache on FIELD_DEFINITION

Implement resolvers:

typescript
// node/resolvers/reviews.ts
import type { ServiceContext } from '@vtex/api'
import type { Clients } from '../clients'

type Context = ServiceContext<Clients>

export const queries = {
  reviews: async (
    _root: unknown,
    args: { productId: string; limit?: number; offset?: number },
    ctx: Context
  ) => {
    const { productId, limit = 10, offset = 0 } = args
    const reviews = await ctx.clients.masterdata.searchDocuments<Review>({
      dataEntity: 'reviews',
      fields: ['id', 'productId', 'author', 'rating', 'title', 'text', 'createdAt', 'approved'],
      where: `productId=${productId} AND approved=true`,
      pagination: { page: Math.floor(offset / limit) + 1, pageSize: limit },
      schema: 'review-schema-v1',
    })

    return {
      data: reviews,
      total: reviews.length,
      hasMore: reviews.length === limit,
    }
  },

  review: async (
    _root: unknown,
    args: { id: string },
    ctx: Context
  ) => {
    return ctx.clients.masterdata.getDocument<Review>({
      dataEntity: 'reviews',
      id: args.id,
      fields: ['id', 'productId', 'author', 'rating', 'title', 'text', 'createdAt', 'approved'],
    })
  },
}

export const mutations = {
  createReview: async (
    _root: unknown,
    args: { input: ReviewInput },
    ctx: Context
  ) => {
    const { input } = args

    const documentResponse = await ctx.clients.masterdata.createDocument({
      dataEntity: 'reviews',
      fields: {
        ...input,
        author: ctx.vtex.storeUserEmail ?? 'anonymous',
        approved: false,
        createdAt: new Date().toISOString(),
      },
      schema: 'review-schema-v1',
    })

    return ctx.clients.masterdata.getDocument<Review>({
      dataEntity: 'reviews',
      id: documentResponse.DocumentId,
      fields: ['id', 'productId', 'author', 'rating', 'title', 'text', 'createdAt', 'approved'],
    })
  },

  deleteReview: async (
    _root: unknown,
    args: { id: string },
    ctx: Context
  ) => {
    await ctx.clients.masterdata.deleteDocument({
      dataEntity: 'reviews',
      id: args.id,
    })

    return true
  },
}

Wire resolvers into the Service:

typescript
// node/index.ts
import type { ParamsContext, RecorderState } from '@vtex/api'
import { Service } from '@vtex/api'

import { Clients } from './clients'
import { queries, mutations } from './resolvers/reviews'

export default new Service<Clients, RecorderState, ParamsContext>({
  clients: {
    implementation: Clients,
    options: {
      default: {
        retries: 2,
        timeout: 5000,
      },
    },
  },
  graphql: {
    resolvers: {
      Query: queries,
      Mutation: mutations,
    },
  },
})

Testing the GraphQL API after linking:

graphql
query GetReviews {
  reviews(productId: "12345", limit: 5) {
    data {
      id
      author
      rating
      title
      text
      createdAt
    }
    total
    hasMore
  }
}

mutation CreateReview {
  createReview(input: {
    productId: "12345"
    rating: 5
    title: "Excellent product"
    text: "Really happy with this purchase."
  }) {
    id
    author
    createdAt
  }
}

Common failure modes

  • Defining resolvers without matching schema fields: The GraphQL runtime only exposes fields defined in the schema. Resolvers without matching fields are silently ignored. Conversely, schema fields without resolvers return null. Always define the schema first, then implement matching resolvers with identical names.
  • Querying external APIs directly in resolvers: Using fetch() or axios bypasses the @vtex/api client system, losing caching, retries, metrics, and authentication. Always use ctx.clients in resolvers.
  • Missing @auth on mutation endpoints: Without @auth, any anonymous user can call the mutation — a critical security vulnerability. Always add @auth to mutations and queries returning sensitive data.
  • Missing @cacheControl on public queries: Every request hits the resolver without caching, causing unnecessary load and slow responses. Add appropriate cache directives to all public Query fields.

Review checklist

  • [ ] Is the graphql builder declared in manifest.json?
  • [ ] Do all public Query fields have @cacheControl with appropriate scope and maxAge?
  • [ ] Do all Mutations and sensitive Queries have @auth?
  • [ ] Do resolver function keys exactly match schema field names?
  • [ ] Are resolvers using ctx.clients for data access (no raw HTTP calls)?
  • [ ] Are directive declarations present in directives.graphql?
  • [ ] Is the resolver wired into the Service entry point under graphql.resolvers?

Related skills

Reference

from this repository

More skills

All skills
vtex
Community

architecture-well-architected-commerce

Apply when scoping, reviewing, or documenting cross-cutting VTEX commerce architecture across storefront, IO, headless, marketplace, payments, or any other VTEX module. Grounds work in the Well-Architected Commerce framework—Technical Foundation (reliability, trust, integrity; security, infrastructure, compliance), Future-proof (innovation, simplicity, efficiency; scalable and adaptable solutions), and Operational Excellence (accuracy, accountability, data-driven improvement; process and customer experience). Routes implementation detail to product tracks (IO caching and paths, Master Data strategy, marketplace integrations). Use for solution design, architecture reviews, and RFP-level technical structure.

installs
2
GitHub stars
42
Updated
Aug 19
vtex
Community

faststore-storefront

Core coding rules and workflow for developing VTEX FastStore storefronts. Use when starting any FastStore development task, writing TypeScript/React components, creating section overrides, extending the BFF, or styling. Covers all primary conventions, safety rules, and the development workflow used across every FastStore project.

installs
2
GitHub stars
42
Updated
Aug 19
vtex
Community

headless-bff-architecture

Apply when designing or modifying a BFF (Backend-for-Frontend) layer, middleware, or API proxy for a headless VTEX storefront. Covers BFF middleware architecture, public vs private API classification, VtexIdclientAutCookie management, API key protection, and secure request proxying. Use for any headless commerce project that must never expose VTEXAPPKEY or call private VTEX APIs from the browser.

installs
2
GitHub stars
42
Updated
Aug 19
vtex
Community

headless-caching-strategy

Apply when implementing caching logic, CDN configuration, or performance optimization for a headless VTEX storefront. Covers which VTEX APIs can be cached (Intelligent Search, Catalog) versus which must never be cached (Checkout, Profile, OMS), stale-while-revalidate patterns, cache invalidation, and BFF-level caching. Use for any headless project that needs TTL rules and caching strategy guidance.

installs
2
GitHub stars
42
Updated
Aug 19