IDs and Paths

Every document in a Sanity dataset must have an ID that identifies it, an arbitrary string of maximum 128 characters made up of the characters a-zA-Z0-9._-. Note that an ID cannot start with a - (dash) character, and must not have more than one consecutive . character. E.g. -abcde-12345 would be an invalid ID, as would -a..bcde-12345.

The ID is specified in the document’s _id property, and must be unique within the dataset. The ID cannot be modified once a document is created, since it is used to track the document’s history and relations.

The Sanity Studio automatically generates a random UUID for new documents (e.g. 189bc292-e41b-42a0-91b5-bfaa33a34af2), and does not allow you to specify an ID yourself.

IDs with multiple segments separated by periods must not include the segment versions unless it is the first segment. When the first segment is versions the ID must have at least three segments. For example, versions.abc.xyz is permitted, but versions.abc and abc.versions.bar are not. These restrictions are due to internal implementation requirements.

IDs are also considered paths, separated by periods.

Sanity also uses paths for storing various internal data in your datasets. For example, internal objects like groups are stored under the _. path, and the content studio stores draft documents under the drafts. path and Content Release version documents under the versions. path.

GROQ provides a path() function that allows you to filter documents by path, such as fetching all drafts with _id in path("drafts.*") or fetching all versions with _id in path("versions.**"). In path expressions, * is taken to mean “anything up to the next period”, while ** means “anything including periods”. The path() function currently only works with the _id attribute, since it requires special indexing.

To work with drafts, versions, and document ID paths we recommend using the @sanity/id-utils helper library.

The Content Lake and Studio automatically generate unique identifiers for documents. These system-generated _id values are designed to ensure consistency and prevent conflicts across your dataset.
There is no way to override the default ID logic, but you can theoretically set custom ID strings for documents created via our APIs or the client.

While it might be tempting to create custom IDs for documents, we recommend a more flexible approach, which offers several key advantages and generally scales better.

Instead of attempting to override the native ID system, create a custom field in your schema to store your preferred identifier.

// product.ts
import { defineType, defineField } from 'sanity'

export default defineType({
  name: 'product',
  title: 'Product',
  type: 'document',
  fields: [
    defineField({
      name: 'customId',
      title: 'Custom Identifier',
      type: 'string',
      initialValue: () => yourCustomIdGenerator()
    })
  ]
})

By following these guidelines, you can effectively manage document identifiers while maintaining flexibility and adhering to Sanity.io's best practices.

If you need to generate IDs in a script or function which are compatible with Sanity's system, you can use uuid() from the @sanity/uuid package.

import {uuid} from '@sanity/uuid'

// Generate a unique ID compatible with Sanity's system
const newDocumentId = uuid()