Feathers and Zod services

A service is the primary functional unit of nuxt-feathers-zod. The CLI generates the structure expected by service discovery and schema tooling.

Standard service

bunx nuxt-feathers-zod add service articles --database mongodb --connection primary --collection articles --schema zod

Starting with 6.7.43, the recommended form is database engine + named connection: --database selects mongodb, postgresql, mysql, mariadb, or sqlite, and the CLI selects the compatible mongodb or knex adapter. --connection then binds the generated service to the NFZ registry.

The existing --adapter mongodb|knex syntax remains compatible, while --adapter memory stays available for in-memory services. If both selectors are present, NFZ rejects incompatible combinations.

For SQL, use --table and optionally --schemaName:

bunx nuxt-feathers-zod@6.8.0 add service audit-events \
  --database postgresql \
  --connection reporting \
  --table audit_events \
  --schemaName reporting \
  --schema zod

The .nfz/services/<service>.json manifest records connectionName, databaseType, databaseProvider, and databaseFamily. schema <service> --show surfaces that identity metadata, and doctor compares generated bindings with named connections from nuxt.config.ts when they can be resolved statically.

Choose an identifier strategy

Starting with 6.7.44, --idStrategy makes the identifier contract explicit without silently deriving semantics from the database engine:

Adapter	Default	Available strategies
MongoDB	objectid	objectid, uuid, string
Knex / SQL	integer	integer, bigint, uuid, string
Memory	integer	integer, uuid, string

PostgreSQL UUID example:

bunx nuxt-feathers-zod@6.8.0 add service api-keys \
  --database postgresql \
  --connection reporting \
  --table api_keys \
  --schema zod \
  --idStrategy uuid

The manifest then records idStrategy: "uuid", and the generated class passes id: "id" explicitly to the Feathers adapter. uuid and string are treated as client-assigned identifiers in generated create schemas; objectid, integer, and bigint are omitted from create schemas by default so the adapter or database can generate them.

bigint is represented at the API boundary as a decimal string ("9223372036854775807"), not as JavaScript bigint. This avoids JSON serialization loss or failures; actual SQL-driver behavior remains subject to the later per-engine certification patches.

Zod query handling preserves the distinction: query-string values are coerced to numbers only for number-backed fields, while UUID, string, and decimal-bigint identifiers remain strings. Textual $sort values "1" and "-1" are normalized to the Feathers sort orders 1 and -1.

Schema modes are none, zod, and json.

Custom service

bunx nuxt-feathers-zod add custom-service reports --methods find --customMethods run --schema zod

find is a standard Feathers method. run is an explicitly declared custom method.

Schema fields

bunx nuxt-feathers-zod schema articles --show
bunx nuxt-feathers-zod schema articles --add-field title:string!
bunx nuxt-feathers-zod schema articles --set-field published:boolean=false
bunx nuxt-feathers-zod schema articles --validate

Protected services

bunx nuxt-feathers-zod auth service articles --enabled

Keep authentication and authorization in Feathers hooks instead of duplicating business security in Nitro API routes.

Client call

const articles = useService('articles')
const page = await articles.find({
  query: {
    published: true,
    $limit: 25,
    $sort: { createdAt: -1 },
  },
})

Server call

const rows = await app.service('articles').find({
  query: { $limit: 25 },
  provider: undefined,
})

No loopback HTTP request is required inside the same server runtime.