Environment

• Operating System: Darwin
• Node Version: v24.6.0
• Nuxt Version: 4.0.3
• CLI Version: 3.27.0
• Nitro Version: 2.12.4
• Package Manager: [email protected]
• Builder: -
• User Config: compatibilityDate, devtools, modules, css, content
• Runtime Modules: @nuxt/[email protected], @nuxt/[email protected], @nuxt/[email protected], @nuxt/[email protected], @vueuse/[email protected]
• Build Modules: -

Version

3.6.3

Reproduction

A reproduction is in the discussion here: #3489

In my use case:

export default defineContentConfig({
  collections: {
    people: defineCollection({
      type: "data",
      source: "org/gdc.csv",
      schema: z.object({
        "Work Email": z.string().email(),
      }),
    }),
  },
});

const { data: tmpContent } = await useAsyncData("tmp-content", () => queryCollection("people").all());

[
  {
    "id": "people/org/gdc.csv",
    "Work Email": null,
    "extension": "csv",
    "meta": {
      "path": "/org/gdc",
      "body": [
        ...my csv data is here...
      ],
      "title": "Gdc"
    },
    "stem": "org/gdc",
    "__hash__": "t_hC0vKvEj6NWa3r_goNjnn4BEvwNnHt6Y8WvNiIUHI"
  }
]

Description

CSV data is not returned from queryCollection as it is for markdown. The result is an array of length 1 with a body attribute that has all the data.

The documentation mentions putting this in the config:

export default defineNuxtConfig({
  content: {
    build: {
      csv: {
        json: true,
      },
    },
  },
});

But it seems to have no effect.

The documentation also seems to have errors. The example schema has "name, email, avatar", then has two example csv files, with "id, name, email", and "name, role, avatar".

Additional context

cc @peter-chipello

Logs